Ejemplo - Block Chain

Journal of King Saud University - Computer and Information Sciences 35 (2023) 101801
Contents lists available at ScienceDirect
Journal of King Saud University - Computer and Information

Sciences
journal homepage: www.sciencedirect.com
Enhanced computer vision applications with blockchain: A review of

applications and opportunities
Najmath Ottakath ∗, Abdulla Al-Ali, Somaya Al-Maadeed, Omar Elharrouss, Amr Mohamed
Department of Computer Science and Engineering, Qatar University, Qatar
ARTICLE INFO ABSTRACT
Keywords: Videos and image processing have significantly transformed computer vision, enabling computers to analyse,
Blockchain and manipulate visual data. The proliferation of cameras and IR equipment has facilitated the collection of
Computer Vision valuable information about individuals and their surroundings. These technologies find applications in various
Video surveillance
domains, ranging from biometric entry cards and high-security clearances to surveillance. These applications
Video integrity
form part of the Internet of Things (IoT), forming a centralized network. However, the proliferation of data and
Video and image sharing
Privacy
its sharing brings challenges related to security, privacy, and storage. Interactions with third-party systems may
introduce vulnerabilities. To address these issues, researchers in computer vision have explored the integration
of blockchain technology into various applications. This paper presents a comprehensive survey of blockchain
applications in computer vision, focusing on image and video data sharing, video surveillance, biometrics,
and video integrity protection. The aim is to explore how the blockchain can enhance the security, privacy,
and authentication of them. It also discusses tools and techniques employed at the edge to achieve these
objectives while highlighting opportunities for further improvements. Overall, this review provides insights
into the integration of blockchain and computer vision, advancements, challenges, and future directions in
leveraging image and video data in a blockchain-enabled environment.
1. Introduction matter how secure, with the multitude of encryption methods available
can be easily hacked or corrupted by tampering at the control centre. A
Tamper proof and immutable record, safe and secure are features single failure without countermeasures may damage the whole system.
of distributed ledger technology revolutionizing not just cryptocur- The centralized system can be an easy point of entry to several security
rency but also integrated to a multitude of fields that enable smart vulnerabilities justifying the requirement of a distributed, decentralized
public security, smart health care, smart cities, and further intelligent system (Sunyaev, 2020). Here we harness the advantages of the dis-
systems (Drescher, 2020). tributed ledger technology (DLT) called blockchain which provides a
Due to the massive growth of new Internet of things devices (IoT) decentralized architecture for safety, privacy, and immutability (Sun-
and sensors networks in almost all industries, catering to the need of yaev, 2020). Computer vision applications are enabled by cameras
the now, has caused a rapid rise in inter-connectivity (Shafique et al., and sensors; images or videos are captured and then processed to
2020). Computer vision applications like surveillance, face recogni- make meaningful data based on the application. Intelligent and au-
tion systems, fingerprint recognition systems, etc. has become part of tomated systems are facilitated by artificial intelligence (AI) (Hafiz
this huge interconnected network of the internet of things (Patel and et al., 2023). Machine learning and deep learning further subsets of
Thakkar, 2020). These devices, which are often controlled by a central
AI have enabled better efficiency as well as exponentiated application
station/authority, might be vulnerable in terms of security, privacy,
scenarios in this discipline (Patel and Thakkar, 2020). However, the
and even counterfeiting (Zheng et al., 2018). A centralized control no
∗ Corresponding author.
E-mail addresses: no1912348@qu.edu.qa (N. Ottakath), abdulla.alali@qu.edu.qa (A. Al-Ali), s_alali@qu.edu.qa (S. Al-Maadeed),
elharrouss.omar@gmail.com (O. Elharrouss), amrm@qu.edu.qa (A. Mohamed).
Peer review under responsibility of King Saud University.
https://doi.org/10.1016/j.jksuci.2023.101801
Received 8 June 2023; Received in revised form 4 October 2023; Accepted 13 October 2023
Available online 18 October 2023
1319-1578/© 2023 The Authors. Published by Elsevier B.V. on behalf of King Saud University. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).
N. Ottakath et al. Journal of King Saud University - Computer and Information Sciences 35 (2023) 101801
content captured is very sensitive in terms of privacy laws, theft of processing speed, reduces cost, and forms a non-repudiated network,
data, creating fake images or videos (Mayer and Stamm, 2020), spoof- enabling the integrity of data to be kept safe (Abuhashim and Tan,
ing (Kamble et al., 2020) among many other vulnerabilities. Numerous 2020). This enables access control and contract enforcement (Mohanta
opportunities in identity management, indexing records, content man- et al., 2018), reduces risks, and cuts down on any third-party costs.
agement, privacy management, and several other applications can be With that comes better efficiency of the process involved (Cheng et al.,
accomplished through blockchain with image data. With no existing 2018). This property makes it suitable for user control management,
survey on computer vision applications with blockchain or vice versa, access control, or encryption, enabling applications like safe and secure
this review summarizes the applications of blockchain with computer data sharing, enforcement of automated contracts, etc. providing safety,
vision and how it is used to solve the issues of privacy, security, and security, and privacy for many diverse applications (Haiwu et al., 2018;
centralized control. This paper reviews the latest state of the art in Hughes et al., 2019b).
detail, summarizes their applications, and categorizes them based on The blockchain can be classified based on its architecture and
their main task. mode of access given for validation (Chowdhury et al., 2019). Public
The main contributions of this paper are: blockchain allows the creation and validation of blocks by the public.
Modification of the data is performed using transactions. This creates a
• A comprehensive review of state-of-the-art applications of transparent and open access framework, raising privacy issues (Guegan,
blockchain with computer vision. 2017). Private blockchain on the other hand is restricted, where only
• A review of applications of its architecture, components, limita- authorized parties are allowed to take part in the activities within the
tions and implementation. blockchain. This can be leveraged for computer vision use cases where
• Opportunities for future work and the most common trends in unauthorized entities should not be able to access any transaction of
applications with computer vision are then presented based on activities thereby achieving security and privacy (Guegan, 2017). A
analysing the limitations and opportunities. consortium blockchain is a blockchain where the consensus mechanism
is controlled by a pre-selected set of nodes. Several organizations can
The organization of the paper is as follows: Section 2 provides a
come together to form a consortium or a federated blockchain Sunyaev
background on blockchain and computer vision. Section 3 presents the
(2020).
methodology for article selection. Sections 4–7 review the state of the
With all these advantages, there are also drawbacks in terms of its
art in video surveillance, biometrics, video and image data sharing, and
security risks, computational costs, scalability, high energy consump-
video integrity. Section 8 provides an overview of all the applications
tion, integration with other systems and inter-operability with other
of blockchain with computer vision and the tools used for deployment,
blockchains, and so on. Careful planning needs to be done to choose
as well as some limitations and opportunities. Then we conclude the
the application area of blockchain and mitigate the risks involved.
paper with key highlights of this survey and future direction.
There is an inherent trade-off where you get an immutable, more
secure, tamper-proof, timestamped record but with some modifications
2. Background
required for scalability, computational costs, and latency involved.
Different types of blockchain solve different limitations such as off-
Distributed ledger technology (DLT) is a decentralized database
chain storage for improving scalability, third-party applications for
that holds transactions between entities in a tamper-proof and im-
inter-operability, and so on. These trade-offs should be taken into
mutable ledger, with each entity maintaining a transparent copy of consideration when designing an architecture for a specific application.
the ledger (Hughes et al., 2019a). Blockchain is one such DLT that Pattern recognition, image processing, object identification, object
keeps a record of the transaction validated by several peers based on a recognition, and classification have been broadly termed computer
consensus mechanism (Drescher, 2020). Bitcoin a type of blockchain vision problems. These tasks have applications in several industries
that solved the double-spending problem in a paper authored by a from autonomous driving to robotics vision applications (Feng et al.,
pioneer with the pseudo name Nakamoto and Bitcoin (2008); it brought 2019).
forth a new interest in this technology which led to a rise in appli- Recent developments in intelligent cameras (Rebecq et al., 2019),
cations from security, data storage, and privacy among many others. where the analytics and inference can be performed on embedded chips
To cater to the different needs of applications and control over the can be utilized for faster results. With the development of machine
network, the blockchain developed into public, private, and consortium learning and deep learning techniques, particularly Convolutional Neu-
blockchains (Andreev et al., 2018). ral Networks (CNN), there has been a huge increase in the number of
Due to the increasing use of blockchains in several applications, applications in computer vision that operate with image and video data.
their requirements compared to conventional methods are usually The architecture of a computer vision setup usually involves an
questioned. Conventional methods or a centralized relational database image capturing device and a device that performs the image pro-
model put their trust in a few or a single entity to secure the data. cessing. The image processing device performs the inference on the
Blockchain is incorruptible, and since each peer of the network contains captured data (Gollapudi, 2019). This leads to networks of devices
a copy of the ledger, trust need not be given to a single authority and layering of the hardware components. Fig. 1 illustrates a Wireless
but rather to trustless peers with a complete ledger of the transaction, Sensor Network (WSN) (Peixoto and Costa, 2017) or Internet Of Things
which may include any type of data secured on their drives (Peck, architecture where edge computing or fog computing is enabled (Shi
2017). A blockchain system consists of a set of linked blocks that et al., 2016). The things here are the capturing devices. Typically,
contain transaction data, a private key, a hash, and a nonce secured the data processing and interface work on the edge. Cloud computing
with a public key. Each block is linked to the previous block, which is required for storage and heavy computations which happens on
contains the hash value of the preceding block and a nonce. The time the cloud layer. This necessitates the use of blockchain to safeguard
of the transaction is registered in each block. A block is added by the transfer of photographs and video, to safeguard the privacy with
validation through a consensus mechanism. A transaction in blockchain integrity during information transfer, and to enable authentication and
can have any kind of data stored, however, with limited capacity based authorization of access, to the information in these edge devices (García
on the type of blockchain platform used. The fact that a blockchain hash et al., 2017). Emerging opportunities in blockchain with computer
cannot be modified or changed and that it is cryptographically secured vision, enhance many current applications from video surveillance to
enables the property of security and trust. In addition to that, smart biometric identification.
contracts can be deployed to automate, control access, and execute a The next few sections of this paper will detail the applications of
contract or agreement. Smart contracts play a pivotal role in automated computer vision with blockchain producing a review on the method-
tasks in the blockchain. Automation through smart contracts improves ology, applications, tools used, and problems solved with some of the
2
Fig. 1. Sensor network in surveillance systems.
limitations of those methods. With these, we can infer the guidelines for Relevance to the study was examined from the selected literature
creating applications of Blockchain with Computer Vision. Section 3, and those inadequate were discarded. The main key terms used for
is the methodology of the review which details the process applied this survey were ’Blockchain in computer vision’, ’Blockchain image
for literature selection. Fig. 2 is an overview of the core applications processing’, ’Blockchain video surveillance’, ’Blockchain Biometrics’,
discussed in this review. We review current trending applications that ’Blockchain video integrity’, ’Blockchain content sharing’ e.t.c. The
perform image and/or video processing that use blockchain such as following flowchart in Fig. 3 summarizes the process.
Video surveillance systems being first in Section 4, was found to focus
on security, content protection and providing access control to view 4. Video surveillance systems
and/or modify the video footage. Biometrics in Section 5 on the other
hand focused on better feature extraction, multi modal biometrics and Surveillance and monitoring are essentials for any safe and secure
fusion methods for encryption of private keys in blockchain as well community. It enables prevention of crimes as well as reduces the dam-
as storing biometric hashes in the blockchain enabling access control age that may be caused due to late response. A perceptive view of the
using identity management using authentication, and security. Video happenings in a surrounding can enhance the action taken (Tsakanikas
integrity were taken as point of focus in Section 6 where preserva- and Dagiuklas, 2018). Smart surveillance through AI has achieved this
tion of integrity of videos through identifying fake images and/or objective (Sreenu and Durai, 2019). Cameras captures multitude of
videos, archiving for preserving history and content verification was images that can be viewed and stored as well as accessed for future use.
performed. Video and/or image sharing in Section 7 explores the state However, safety, privacy, and vulnerability caused by unauthorized
of art on secured sharing, medical image sharing and video streaming access is a big concern.
applications, functioning with a blockchain or distributed layer. In a typical surveillance system, multiple surveillance devices cap-
Many concepts and architectures are charted out, however there is ture the videos. These videos are then processed on a single central
need to implement this in practice. The evaluation metrics for each vary server, or a control station. This forms an edge/fog network kind of
on the application, however there is a need for specificity for measuring architecture as illustrated in Fig. 1. The data is eventually stored in
performance. Nevertheless, it can be proved that images can be secured, the cloud or sometimes even processed there. However, the Internet of
users made accountable, time stamped and uneditable through use of things paradigm given to the multi-surveillance systems brings forth a
blockchain. Most of the applications are based on public security where lot of security and surveillance risks (Wang et al., 2019).
images or videos of people or property are bound by laws. Blockchain Centralized storage has a single point of entry which when infil-
can now enable secure monitoring of these without the need to over- trated can cause sensitive and private image data to be accessed as
come these limitations in terms of access. Computer vision process well as tampered with (Mittal et al., 2020). Even though cryptographic
can further be enhanced with distributed learning opportunities, secure mechanisms and security tools are available to share or store the
encryption of training data as well as the trained weights. Sharing has data, any vulnerability in them can be taken advantage of, and can
become more secure and safe within limits of the privacy laws. bring down the whole system resulting in the loss or damage of it.
Blockchain provides a decentralized storage as well as mechanisms to
3. Methodology control access to the stored information. Adding to that blockchain can
safeguard privacy and keep away from tampering.
The aim of this research is to identify applications related to com- In the following section we review the applications of video surveil-
puter vision and image processing using blockchain in several domains. lance system using blockchain for content protection, security and
For this purpose, the literature was surveyed on topics related to access control enabled systems.
blockchain and possible applications such as video surveillance, bio-
metrics, video integrity preservation and video/image sharing. Further 4.1. Content protection
survey was performed on how blockchain is involved in each and
what are its advantages and limitations. To conduct this study promi- Video and image data that are captured, often is sensitive infor-
nent databases were surveyed, such as Scopus, IEEE explore, Google mation. Content generated or captured need to be protected with
Scholar, and Web of Science. Duplicates in each were eliminated. respect to laws related to privacy and individual freedom of choice
3
Fig. 2. Overview of common computer vision applications with blockchain.
if the individual is not a significant party in the video. Unauthorized Video and/or image content can be protected by utilizing feature
access and tampering of video/image are some of the major flaws in of the blockchain where modification is monitored as any transaction
a centralized system. Available devices in a centralized system are not is time stamped. Fitwi et al. in Fitwi et al. (2019), used lightweight
equipped to monitor and be selective within a wide range of coverage. blockchain named Lib-Pri for privacy protection where tasks like check-
ing integrity of the videos, blurring keys management, feature sharing
A scheme to protect the content as well as privacy of those involved
and video access sanctioning was performed. Edge computing was
are the main focus in the following state of the art where blockchain is performed for real time video analytics in this paper for identifying
used to mitigate this. Some of the implementations also point out the suspected individuals (Fitwi et al., 2019). Video is split into frames
requirement of a scalable and light weight blockchain Chauhan et al. and a reverse chaotic mask is applied to images which is then stored
(2018). in an off-blockchain storage. The suspected individual’s facial features
4
entities were given tokens based on privilege of access. The application

layer of the video chain updates and verifies the evidence. A new data
storage mechanism was built called Trusted Video Evidence Storage
(TVES); it stores both evidence and original data (Liu et al., 2018).
A high transaction rate protocol the VideoChain consensus protocol
was used for validation. In the video processing part, the evidence was
collected by cutting it 10 min apart and then compressed. A hash of
the video was computed as evidence of video integrity. Reliability is
thereby improved by adding backup to the original video. Analysed
based on security and efficiency video chain proved to be a suitable
option for implementation in real time.
A blockchain network was setup through several application do-
mains such as, Unmanned Aerial Vehicle (UAV) and dam surveillance.
It was utilized for security of data stored and accessed. The authors
in Youssef et al. (2019) devised a secure IoT solution for dam surveil-
lance where a distributed and long term security solution is accom-
plished through blockchain technology by providing authentication,
data storage, integrity and traceability of data delivery through the
UAV cloud. The performance was measured based on the data delivery
ratio.
A real time surveillance requires not just video storage but also an
intelligent and most relevant video frames or images to be stored which
seemed to enhance the inference and fast action for surveillance ap-
plications. A combination of CNN, Inter-Planetary File Storage system
(IPFS), edge computing and a permissioned blockchain was utilized
for massive data storage, real-time monitoring and large scale infor-
mation acquisition (Wang et al., 2019). Content oriented surveillance
was accomplished by identifying dangerous individuals using sensors
to identify malicious activity and tracking them after identification.
Fig. 3. Article selection process.
Passive imaging and detecting concealed objects were performed and
secured where data integrity was kept valid using blockchain (Qi et al.,
2020).
are compared with the original and pushed to the blockchain node to Multilayered network is usually used with a private blockchain
identify the location and identity. Thus preserving the privacy of those for secure surveillance in addition to a secure storage. The term
involved. IBSS(independent blockchain surveillance system) is introduced in
In the same context, camera identity stored as transaction data Singh (2020) where high level layers formed the block chain layer in
on the blockchain was used to preserve the privacy of individuals which the data is hashed and stored in an IPFS storage. The sensors
in Gallo et al. (2018). The authors in paper designed Blocksee which is (cameras) act as nodes to the blockchain where the video content is
a video surveillance system in smart cities where vulnerability caused hashed and saved to the IPFS which is validated through a consensus
by malicious users manipulating the video content was solved. Camera mechanism. With the IPFS system, a secure large data storage was
identity enabled validation and immutability where camera settings achieved with blockchain preserving privacy and security of the video
were stored, which contained camera ownership details enabling pri- data captured from the sensors. Authors in Nyaletey et al. (2019) used
vacy through blockchain. Preserving the camera details safe-guarded an IPFS system and to secure this peer to peer network a block chain
the content and enhanced privacy of the images captured by the owner is proposed.
of the camera. Further authorization to access to the images were also
controlled in this process. 4.3. Access control
A case of privacy and security for a person of interest detected on
camera in surveillance systems was taken into consideration by authors Controlled access to the blockchain transaction secures privacy of
in Fitwi and Chen (2021). Video sharing was performed using a pri- the individuals as well as restricts the access to unauthorized individ-
vate permissioned blockchain authorized using smart contracts. Video uals or organizations to use images or videos of entities not involved
frames were enciphered using a discrete cosine transform and advanced in the surveillance incident (Gallo et al., 2018; Fitwi and Chen, 2021).
encryption mechanisms. This enhanced layered security enabled better Jeong et al. in Jeong et al. (2019) surveys blockchain based manage-
security and privacy preservation along with access control. ment of surveillance systems and Deepak et al. (2020) surveys the use
of hyperledger fabric (Vukolić), a type of private blockchain managed
4.2. Security with IPFS and CDN in surveillance applications. In Islam and Shin
(2019), to secure the data acquisition during an UAV surveillance, the
Security is the key factor in credibility of data generated and shared data was stored in a blockchain at the mobile edge computing server.
in surveillance systems. Blockchains provide a secure means of sharing Storing surveillance footage and controlling access can be per-
and storing data through encryption and the use of hashing algorithms. formed by blockchain smart contracts. Bálint (2020) introduced a
Credibility of video content as well as its integrity are preserved. blockchain based system for storage of video footage where data was
Evidence gathering through footage stored in blockchain where stored off-chain, where storj platform for storage was found to have
its feature of timestamped storage and immutable record is taken an elaborate encryption method which secures the data along with
advantage of in several applications. Video credibility was the point storing it. Several types of decentralized storage were also compared
of focus in Liu et al. (2018) where the authors used Video-Chain which in this paper for use as off-chain storage for blockchain. Nikouei et al.
is a blockchain for video evidence storage. Video integrity evidence is in Nikouei et al. (2018) identified the problem of collecting data at the
saved on the blockchain. It follows a consortium blockchain, where the edge along with feature extraction. The closest fog nodes then classify
5
Fig. 4. Overview of surveillance system with blockchain.
the features. Misleading the surveillance system can be easily done at through smart contracts or chain codes in this application. Scalability
multiple layers where tampering can happen in the cloud system. A was achieved due to the consortium architecture. Real-time suspect
blockchain enabled scheme was applied to protect the index of the data monitoring was achieved successfully with minimal delay. Access con-
stored at the cloud, at the edge and fog, to secure the data and smart trol through smart contract enables restricted access to private and
contracts can be used to authorize access to that data. confidential information shared through different organization with
Considering the requirements of a real time model, Singh et al. integrity and confidentiality.
in Singh et al. (2020) created a lightweight mechanism named ODOB,
which was used for surveillance with drones. ODOB decouples the block 4.4. Discussion of video surveillance systems
ledger from the block header to form a distributed architecture. Here
each drone can access only their own block and this makes it simple, With safety and security preserved on the blockchain for surveil-
trustworthy, and lightweight. Another lightweight implementation of lance and with a general overview of the systems used, Fig. 4, illustrates
blockchain with surveillance was achieved in Michelin et al. (2020) a general architecture of block chain in surveillance. The video surveil-
where video metadata was stored on the blockchain to support video lance system can be part of an edge network or sensor network. The
integrity. The Video metadata such as frame rate, video position se- sensor data (video and/or image captured) is typically stored in a
quence, video frame and Storage address was stored as transaction. The central station at the edge of a wireless sensor network for further
video was stored in a distributed storage system called IPFS. The setup processing which can be treated as a blockchain node (Nagothu et al.,
2018). Fig. 5 contains the key components in the surveillance system
of the device included a video camera module connected to a raspberry
with blockchain with a modular architecture of a blockchain based
pi3 with IPFS storage running on a private instance. After evaluating
system.
for latency it was found to only take 8 ms for the whole process from
The Central station where the image and/or video processing is
capturing to storage in IPFS system. Thus achieving video integrity in
done can be considered as nodes to the blockchain network which
surveillance.
stores the video hashes either as a transaction or onto a secure
Apart from the human aspect of suspicious behaviour, Farr et al.
blockchain based storage such as IPFS (Nyaletey et al., 2019). Most
in Farr et al. (2020) used surveillance for autonomous detection of
surveillance systems were found to be private or consortium blockchain
stolen car and inspections where a blockchain based platform was
to control access. Access can be coded with a smart contract re-
used for verification. To identify suspicious vehicles, camera feed was
stricting access to different users of the system enabling privacy for
processed both remotely and locally. Verification of the suspected was the required parties. Thus blockchain enables security, safety and
achieved by blockchain validation system. Anomaly was determined privacy (Destefanis et al., 2018).
by an open source licence plate recognition model—DeepANPR (Deep As with every application in blockchain, tamper proof record and
learning based Automatic number-plate recognition) using the SUN immutable record is guaranteed. The surveillance system is prone
database for licence plate detection. Car recognition was achieved to many video integrity and privacy issues that can be solved by
using a ResNet-152 architecture and trained using Cars dataset (Ke blockchain. Access control which can be programmed through smart
and Zhang, 2020). Once an anomaly was detected and verified a local contracts using private and consortium model blockchains can enhance
authority can be alerted and published in the blockchain. This system privacy of those involved. Smart contracts were applied to the block
ran at a fault free time of 40 s producing fast results (Farr et al., 2020). chain for controlling the image retrieval or viewing capability of dif-
With delay and latency taken into account, Al-Sahan et al. (2020) ferent entities enabling privacy and security of the video data. They
utilized hyperledger fabric as a private permissioned blockchain for were programmed to validate or to verify the authority of the entities
public security. Access control was accomplished through chain code in the blockchain for image and/or video viewing and image and/or
or smart contract. Machine learning and blockchain were combined video retrieval. Encryption of the videos used in surveillance further
in this method where real time surveillance was used to identify improves safety of the videos.
and track suspected faces. Here the latency was used as evaluation Table 1, lists out the existing literature on video surveillance system
metric to identify its effectiveness in reporting a suspect. Surveillance with blockchain, The use of each are listed with the computer vision
events were notified and embedded on to the permissioned blockchain, problem solved and the type of blockchain used to secure and provide
specifically hyperledger fabric which further enabled access control privacy and access control.
6
Fig. 5. The key components of the blockchain system with video surveillance including the modified blockchain structure, the infrastructure and the image/video processing
involved.
It can be clearly noted that most of the applications that were features (Bolle et al., 2013). A biometric system typically consists of a
integrated for national security and general suspicious activity or object biometric device used to extract the biometric features of a person that
identification, used a consortium blockchain or private permissioned are unique like that of physical features as well as behavioural features
blockchain. Here access control is administered using smart contract. of a person (Jain et al., 2011). These features are used as templates then
Image processing or video processing tasks are mostly done indepen- matched during the process for authentication or verification tasks. The
dent of the blockchain at a separate layer. Blockchain is used to provide physical features of a person may include, face, palm, iris, finger print
privacy, security and with that, access control to the already processed and hand veins whereas the behavioural features may include the gait,
or stored data. Most applications, stored the video data on distributed voice, signature, key stroke dynamics etc. of a person (Zhang et al.,
storage systems such as Storj and IPFS. On other hand, lightweight 2011; Elharrouss et al., 2020; Jain et al., 2011).
applications hashed metadata and in some cases video frame of interest, The following section details the applications of biometrics with
and stored as transaction on the blockchain. This resulted in an efficient blockchain with security and access control as the main point of focus.
blockchain system with lesser latency compared to others as on-chain
storage was enabled. 5.1. Security
The limitations of the blockchain mimic the limitation of blockchain
with surveillance. Scalability issues, storage issues, cost depending on Like any image, biometric images can be modified and meddled
type of blockchain, are some of the limitations of blockchain Wang with, thus it needs to be secure. Biometrics like iris and fingerprint
et al. (2019). The efficiency of most video surveillance models rely can undergo spoofing attacks (Hadid et al., 2015). The risks involved
on real time processing. Latency is vital for this role. With a central- in biometric can even lead to fake identities, thefts, impersonations
ized system and single storage multiple access system, latency can be and frauds. Using blockchain with biometrics for several applications,
improved risking privacy and security. However, the sensitivity of the provide that integrity by having an immutable and trustable record.
applications reiterates the need for blockchain. The question arises, can A biometric authentication system that is traditionally centralized,
the blockchain model be enhanced for latency through modification of may be prone to attack and can extract information to steal identities
the block. Some states of arts have brought forth modification of the of people. Most systems involve third party identity providers. They
blockchain block. How far can the model be modified so that there is broker authentication protocols which is one of the main vulnerabilities
significant change in latency and transaction time with the amount of leading to loss of personal information and biometric data (Jain et al.,
data that is to be stored and processed. Edge computing and storing 2008).
data on the edge as a node of the blockchain is still a research direction Authors in Reddy et al. (2020) extracted face and body expressions
in its prime. using a 3D CNN algorithm, which were stored in blockchain called
BigchainDB which is a blockchain for big data. Information extraction
5. Biometrics was performed by detecting the faces, recognizing faces and object re-
identification. 3D CNN used here improved the identification of moving
Biometrics is a person’s unique identity; numerous elements of a objects where spatial and temporal features were both determined from
person, such as their physical and, in some situations, behavioural the video surveillance system. Face and body expression features were
identities, can be unique, allowing a person to be identified and rec- extracted from a pre-processed video sequence. These feature contain-
ognized (Jain et al., 2007). Biometrics has its applications in several ing objects (person of interest) were then tracked and the behaviour of
fields which involve identity management, public security and other the person of interest’s gestures are identified. These features if relevant
applications that need an unique identifier using human biological were hashed and stored in the blocks as transaction data which also
7
Table 1
Video surveillance system with blockchain applications.
Method Application Technique Blockchain type Simulation Smart contract Limitations
Al-Sahan et al. (2020) National security Suspicious face Consortium Hyperledger Yes Camera has to be at a
detection fixed angle to capture
frontal face images. System
requires security testing.
Scalability issues
Youssef et al. (2019) UAV based Dam Fault detection Hybrid Bitcoin No Highly complex system
surveillance with proof of Work
consensus which causes
high energy consumption
Fitwi et al. (2019) Identification of Frame splitting , Private lite-federated blockchain Yes off-chain storage
suspicious Object vulnerabilities
activity,
detection and
recognition
Farr et al. (2020) Vehicle detection Licence plate Public Modified block No Miners are and uses a
and car cloud based DMV)
recognition
Qi et al. (2020) Multi- Person – – Yes Blockchain end to end
surveillance identification evaluation is not provided
sensors and tracking,
object detection
Wang et al. (2019) Suspect Monitoring using Private Permissioned Yes Scalability and security is
identification CNN not tested
Singh (2020) Multi- Video storage Private Permissioned Yes Requires off-chain storage,
surveillance cost and scalability is not
network measured
Michelin et al. (2020) Person Video streaming Private Speedy Chain – Requires off-chain storage
surveillance and and splitting
forensics
Liu et al. (2018) Video evidence Integrity Private videochain Yes Not considered processing
gathering requirements of the video
Lopes et al. (2019) Robot Person Consortium Robot chain Yes Complexity analysis of the
monitoring identification face recognition is required
Gallo et al. (2018) Smart city Segmentation Private Permissioned Yes scalability of the approach
monitoring and feature is not analysed
extraction
Nagothu et al. (2018) Smart micro-service Private Permissioned yes Security measures and
surveillance architecture computational overhead is
not discussed
Islam and Kundu (2018) Indoor Preserving Public – Yes Evaluation of the system is
surveillance IP privacy not performed
cam
Lee and Park (2020) CCTV Preserving – – – Transaction speed and size
surveillance privacy required to be quantified
for merkle tree approach
Bálint (2020) Video footage Encryption or – – – Employs complex
hashing of video consensus mechanisms
footage such as proof of work and
proof of stake
Fitwi and Chen (2021) video storage Preserving Private Permissioned Yes Requires evaluation of
privacy and transaction time while
security network scales
contains the timestamp as well as the location of incident. A chain of which incorporates all features of a typical blockchain where data can
blocks was formed with several blocks containing these features (Reddy be stored both off-chain and on-chain McConaghy et al. (2016) .
et al., 2020). Self-sovereign identity is a concept that brings privacy and security
In the blockchain network a separate copy of each transaction is in identity management in the decentralized systems. The users are
responsible for their own identity data. Blockchain technology is em-
kept and each block contains the block of the previous. This makes it
ployed in this system to establish a web of trust, creating a resilient and
harder for any tampering of data by unauthorized personnel. And so
secure storage environment for identity information. This information
the face and gesture information is secured in the blockchain. Proof
is stored in the blockchain, eliminating the need for a vulnerable central
of work (PoW) was identified as the preferred consensus mechanism database susceptible to hacking. The identity is stored in blocks which
where the nodes in the blockchain, compete to find a nonce value in accumulates the transactions created by many devices. In Othman
the blockchain to produce a hash value. The node that finds the hash et al., A horcrux protocol, a decentralized authentication method for
value creates a block which was approved by the consensus which self-sovereignty of identification using biometric credentials, elimi-
accepts the biggest chain with largest total problem increasing the nates the need of third party identity providers for authentication in
blockchain network. The BigchainDB here is a distributed database blockchain technologies (Othman and Callahan, 2018).
8
Fig. 6. Typical biometric blockchain architecture for identity security with three nodes as peers of a blockchain network. Biometric hash is stored in each peer as a block
transaction.
secret scheme which are then stored using a multi-distribution scheme.

Blockchain was then utilized to store the reference of the decentralized
storage. This scheme allows portability, inter-operability and protection
with that self-sovereignty over a persons own biometric information.
Full control can be achieved by the owner to perform tasks such as
editing, modifying and providing access.
Encryption techniques are prevalent in biometric security. Iovane
et al. in Iovane et al. (2018) presented an encryption system us-
ing information fusion with face biometrics and prime numbers for
public key cryptography and face biometrics for digital currency ex-
changes. The fusion was performed using a face information fusion
algorithm (FIF) also called bio-hashing. This process is employed in
several non-blockchain applications like palm hash, digital finger prints
and face fusion which are later fused with pseudo random codes (Lu-
mini and Nanni, 2006). Here face recognition and finger print recogni-
tion was the main technique used. Smart contracts were employed in
blockchain as self executing contracts and rely on the users in the peer
to peer system. An authentication system is required for the purpose
of identification of the user especially in a private blockchain like the
hyperledger.
Storing biometric templates in a blockchain was presented in
Delgado-Mohatar et al. (2019), where face and hand written signa-
tures were stored after merkle hashing. The face and hand written
features were extracted from a pretrained model. Ethereum architecture
equipped with smart contracts developed in solidity was used for
this purpose. Delgado et al. experimented on several techniques in
blockchain storage such as direct and hash based. It was identified that
Fig. 7. The block structure of a modified block for blockchain system with biometrics merkle tree hashing was more efficient in terms of cost performance
and vice versa along with common biometric profiles used with blockchain in state of
trade-off.
art.
Biometric for security can be elaborately used for several applica-
tions. Mohsin et al. in Mohsin et al. (2019) used a hybrid, biometric,
steganography, blockchain model to secure medical data. Steganog-
The horcrux protocol is a protocol that works with the 2410-2017
raphy is the process of hiding data inside another data (Hassaballah
biometric open protocol standard (BOPS) (Othman and Callahan, 2018)
et al., 2020a; Hameed et al., 2023; Hassaballah et al., 2020b, 2018;
which contains a client device, a trusted BOPS server, and an intrusion Hameed et al., 2023). Finger vein (FV) biometrics which is contactless
detection system for biometrics systems. BOPS can be a single or multi- biometrics system can be applied for access control, authentication and
distribution model having a secret scheme. The multi-distribution electronic passport application. The finger vein extraction technique
scheme enables the storage of biometric data, which can be accessed works by acquiring data through contactless means such as infra-red
through verification using the Horcrux protocol and a biometric au- imaging. This image data is processed, and then matched to existing
thentication process. The data was divided into n templates using a finger vein database for person identification. A verification framework
9
for patient authentication was designed and implemented for the infor- secret key was generated from users’ biometric and then erased after it
mation transfer between the access point and the database where the was used. The keys were thus not saved enhancing safety of the key.
information is stored. Here the risk of loss of key can be reduced. The overhead was found
A combination of RFID and FV biometric was used using a merge to be reasonable in this architecture. However, automatic transactions
algorithm to create a hybrid biometric pattern model which increased such as smart contracts were not supported in this implementation.
randomization and security. Then a combination of blockchain, Immigration and border control are applications where biometric with
steganography, encryption and blockchain was used for transporting blockchain can be used as stated by Patel et al. in Patel et al. (2018).
the biometric data from image acquisition side to the database. After Gateless entry was achieved with hyperledger fabric architecture. The
hybrid and random binary patterns were generated by merging, the decentralized architecture offers security and privacy by enabling con-
features were stored on the blockchain with particle swarm optimiza- trol over document access, determining who has permission to access
tion (PSO), steganography and advanced encryption technique (AES) the documents. Biometric storage can be safeguarded by this approach.
encryption. The node of the blockchain contains an immutable copy of Several encryption techniques were used to improve the safety and
the data with a ledger of user hash. While any user access to a patients security of the biometric information. Toutara et al. in Toutara and
documents is required to be authenticated, reverse of all the procedures Spathoulas (2020) presented a biometric authentication scheme for
can also be performed. This architecture was impervious to brute blockchain where the data from sensors once acquired was homomor-
force attack as well as spoofing attacks, providing, confidentiality and phically encrypted, transformed and then uploaded on to a IPFS system.
integrity. This entire system provides comprehensive security measures, The smart contract also stores the address of the files in IPFS. This
successfully meeting all security requirements. was further connected to the Ethereum address of the user. Ethereum
/IPFS combination was used to achieve privacy as well as security
5.2. Authentication of biometric template. The user additionally saves an encrypted copy
of the transformation parameters and the pair of encryption keys for
For public services, security of national identity is essential, which the template. This mechanism goes through a registration phase where
is a record of a persons biometrics along with other fundamental details the biometric data was acquired and stored. After the registration
of the person. Access to sensitive information needs authentication and phase, authentication can be performed against a third party service
cannot be shared publicly without the required permissions. Blockchain through his/her biometric data. In their experimentation, Toutara et al.
provides secure universal access to biometric information for identity observed that they achieved minimal time overhead.
management for public security and public services (Jacobovitz, 2016).
Blockchain is used in Mudliar et al. (2018) to decentralize the current 5.3. Discussion of biometrics
centralized system for security and transparency for voting applica-
tions. A smart contract was used here to input the data; validated and Biometrics being a unique identification in itself has achieved secu-
extracted data is then stored in blockchain. This information can be rity, however prone to spoofing and theft, blockchain acts as an added
viewed by multiple governmental entities and can be used for several security to any system. With several modes of use of biometrics with
applications like voting identity as performed by authors in Garg et al. blockchain from managing identities to authentication and cryptogra-
(2019). phy, decentralized systems, safety, and security through biometrics can
Breeder documents being a personal profile document containing be achieved.
all essential information about a person are used for authenticating Fig. 6 and Fig. 7, describes a general architecture of biometric stor-
identity of an individual and their credentials. Authors in Buchmann age with blockchain identity with key components of this blockchain
et al. (2017) stored breeder document and electronic travel documents network. Uni-modal or multi-modal biometric data can be stored as
in a decentralized bitcoin blockchain. Breeder documents with biomet- a key alone. It can also be fused with numerology using encrypting
ric information was converted into bar-coded information and with that methods. Biometric hash can be stored on the blockchain system as
personal information was integrated onto a barcode or a chip. This was transactions. The biometric key can also serve as a digital signature.
added on to the documents like birth certificate and national ID. These Table 2, tabulates the applications of biometrics in literature with
documents were further stored on to the blockchain as a transaction. type of biometric used with computer vision techniques used to extract
With the use of image compression on iris and fingerprint images as the biometric features for creating template. The most common biomet-
identifiers, there is significant reduction in the storage space when ric scheme used was multi-modal biometrics. Iris images coupled with
stored onto a stacked 2D barcode. This efficiently identifies and se- fingerprint was the most popular form of biometric. Smart contract is
cures the documents. Biometrics embedded with the breeder document deployed for access control as well as authentication in most of the
provides enhanced recognition and authentication. This authentication applications.
system enables secured access control. Biometrics added authenticity to the blockchain network when
A multi-modal biometric authentication system was another tech- used as an encryption method or as digital signature. It enhanced
nique for this purpose which is incorporated in the blockchain Sawant the security of the blockchain when used as an access key or public
and Bharadi (2020). Sawant et al. suggested a fused iris and face for key. Its use as identity management, made a reliable person identifica-
biometric identification. A CNN architecture was used for classifying tion and recognition system. Security to the biometric templates was
apart from that fusion can be performed by a fully connected layer enhanced when template was stored in the blockchain. Several appli-
with input from CNN layer which extracts the required features. This cations related to national identity, national security, voters identifica-
was planned to be used as a cloud based software. Hyperledger fabric tion, identity documents are main focus for biometrics with blockchain
architecture was suggested for a execute-order-validate architecture application.
which also enables use of smart contracts namely chaincodes. Authen- Finger vein authentication scheme was the most notable of the
tication was achieved by the biometric SaaS (software as a service available approaches for patient document security where the data was
application) containing biometric template. The authenticated peer encrypted by merge algorithm and re-encrypted using steganography
(node) was then allowed to perform transactions based on a consensus and Advanced Encryption standard (AES) which then was stored on the
in the permissioned blockchain. blockchain. This technique was most efficient against known threats
Another method for implementing secret keys for blockchains is as well as for identity management and integrity of the data even
using biometric fuzzy systems (Naganuma et al., 2020). Management of though multiple phases were required to achieve this. However efficient
secret key was used for authenticating the user and the cryptographic in security and safety, it may result in high computational cost and
digital signature, using biometrics based digital signature scheme. A time. There is a trade-off between security and performance overhead
10
Table 2
Biometrics with blockchain application.
Method Application Biometric type Technique Blockchain type Validation Smart contract Limitations
Mudliar et al. (2018) National Multi-modal Feature Public Ethereum No Privacy, Many legal
Identity biometrics extraction, formalities,
Biometric interoperability,
recognition implementation complexity
and infrastructure is not
considered, cost.
Buchmann et al. (2017) Securing Iris and Compression of Public Bitcoin No Bitcoin consensus
identities Fingerprint biometrics, mechanisms complexity,
stored as 2D interoperability, scalability,
barcodes cost
Iovane et al. (2018) Encryption Iris JDL method to Public – Yes Time complexity will add
fuse prime to the latency of the end
numbers and to end system
biometric data
Sawant and Bharadi (2020) Biometric Multi-modal Fusion of Consortium Fabric Yes Storage limitation and
Authentication Biometrics multi-biometric, control within a few
VGG16 on fused organizations, Not network
features for fault-tolerant
classification
Naganuma et al. (2020) Digital Multi-modal Features Consortium Fabric No Storage limitation and
Biometrics extraction control within a few
signature organizations
Li (2020) Distributing Fingerprint Finger print – – – Storage constraints and
images chaotic image scalability
encryption
Delgado-Mohatar et al. (2019) Template Face and hand VGG-face – Ethereum Yes Transparency and
storage images pretrained accessibility of information
written signature to all participants, cost and
performance trade-off
Toutara and Spathoulas (2020) Authentication Any biometric Extract data Public – Yes Limitations of storage with
from sensor, respect to scalability
apply
homomorphic
encryption
of multi-modal biometrics with blockchain where several steps are This approach utilized the Ethereum blockchain to provide a decentral-
involved in a transaction being processed such as hashing, encryption, ized proof of authenticity. IPFS storage, Ethereum name service(ENS)
consensus mechanism, miner validation among a few. and a reputation system were the main components of this system. The
The use of multi-modal biometrics and fusion techniques may in- new source and the artist create the smart contracts. The metadata
crease the overhead of the overall system. Storage capacity may in- requires the artist’s Ethereum address as well as the smart contract
crease with multi-modal systems, along with complex encryption tech- address. This information with video content were stored in an IPFS
niques increasing computational complexity. This although secure may storage which generates a unique hash, addressing the files that contain
need high processing power. There lies a trade-off which needs to be the video on the distributed storage.
addressed based on the priority of the requirements of the applications Smart contract embedded in the Ethereum blockchain contains
it is eventually utilized for. A standardized wholesome model for an access functions to edit, share and distribute based on the terms and
optimum biometric system need to be determined for a safe and secure conditions which was authored by the digital art creators. Any edited
performance withstanding faults like that of spoof attacks and template video of the original video was named as a child video which were
ageing without the complex computation. added on to a list in the smart contract. Thereby enabling tracing of
the edited version of the videos.
6. Video integrity The Ethereum address of an artist and his/her real identity are
correlated with an ENS service, where the artists name, company and
One of the most striking properties of blockchain is data integrity profile matched to their Ethereum address were stored in a decentral-
as it is uneditable and time is stamped upon transaction creation ized fashion. In addition, off-chain resources can be utilized where
(Drescher, 2020). This enables the data to be trustable leading to Ethereum address of a video and the owner can be linked. Artists
many applications in combating fake videos and/or images, cataloguing reputation can also be tracked using a decentralized reputation system.
images for history as well as collecting non tampered evidence. The The profile of the video creator was linked to this system which enabled
following section details the current literature in video integrity preser- reputation to be quantified. Smart contracts are used to add reviews and
vation for videos and images including the processing of the images comments and was then stored onto a decentralized storage system.
prior to blockchain layer. With these feature set, video provenance was secured as well as its
value was calculated based on a reputation system enabling a solution
6.1. Combating fake videos for wholesome integrity protection.
Upon evaluation of this system, it was found that the cost changes
Fake videos being a major cause of concern in the current era, Hasan with the changes in smart contract. Each function executed in the
et al. in Hasan and Salah (2019) used blockchain with which history blockchain has a cost, which includes, transaction and execution
tracking and provenance of videos were safeguarded. The history of charged as gas currency in Ethereum. The security of this system
digital content was traced and tracked using Ethereum smart contracts. encompasses the integrity, accountability, authorization, availability
11
and non repudiation thus making it resilient against several attacks like using API’s. The transaction was processed and verified with the data
Man in the middle, replay and DDoS attacks. stored on the chain. Here Proof of work mining was not required
A similar approach to identifying manipulated videos were by lever- which is computationally expensive thus reducing the computational
aging the uniqueness produced while hashing. Dhiran et al. used this overhead due to mining.
feature where their features were hashed using cryptographic algo- Video captured from a streaming IoT device was used for video
rithms like MD5 and AES and video was stored in the hashed format. forensics where hash of video frames were sent and stored on nodes
Once the tampered video was uploaded a modified hash was generated. of permissioned blockchain in Danko et al. (2019). Connection be-
This can be used to detect a video’s authenticity (Dhiran et al., 2020). tween the video and storage device was established using TCP based
Like Dhiran et al. and Hasan et al., hashing was performed on tool. In terms of the video quality, resolution of videos were a major
the video by Yatskiv et al. in Yatskiv et al. (2019). Video integrity concern. Moderate resolution videos could only be implemented and
can be preserved by protecting the video from unauthorized changes there was a need for further investigation into higher resolution images.
with this approach. Here the hash function of each video frame was Hyperledger fabric was used as the blockchain platform which is a
computed rather than the incident video frame, the hash sum of the permissioned blockchain. The access was restricted to stakeholders
first video frame and the second video frame was consequently added utilizing chaincode, smart contracts for hyperledger blockchain.
up to get hash function. Further this hash was added onto the hash
of the next video frame and so on to form a blockchain transaction. 6.3. Content verification
Video file processing was performed using ffmpeg where video and
audio materials are packaged into container formats. As the hashes are Several implementations like that of Cremona et al. (2019), Chen
dependent on the previous hash function, modification of a frame will et al. (2020a), Yatskiv et al. (2019), Hasan and Salah (2019), Dhiran
change the hash sum which will detect any tampering in the video et al. (2020) and Bui et al. (2019) used content verification apart from
frames. This was evaluated on time spent to generate the hash, which identifying fake images, video evidence or cataloguing and preserving
was found to be proportional to video resolution and the number of videos.
frames selected. With the amount of data prevalent, there is a inherent task to
With the emergence of false news on the internet, Chen et al. verify and analyse them. Video content is very abundant; available
in Chen et al. (2020a) proposed an incentive-aware blockchain-based on several devices like mobile phones, tablets, Personal Computers
solution based on smart contracts and a consensus algorithm tailored (PCs), Closed Circuit TV (CCTV) etc. which are equipped with cameras.
for authority verification. They not only focused on images but also Authors in Zelensky et al. (2018), stressed the need for video content
a combination of types of media involved. This was a preventive verification where the need for data confirmation was a necessity at
approach to fake news propaganda. The customized algorithm used in several public security as well as governmental institutions. A swype
this approach used a form of dynamic weighted ranking evaluation code was used to detect the movement of the camera which was
score. Integrity and location privacy while sharing dash cam videos utilized for verification along with other sensor data of the device. A
were taken as a matter of concern pointed out by authors in Kim et al. decentralized application was implemented named PROVER app where
(2020). An automatic dash-cam sharing system was accomplished using it was initiated when camera was turned ON. Ethereum platform was
deep learning and smart contracts, where the encryption of the accident used with smart contract enabled for verification.
location was performed with that maintaining the location privacy. In order to safeguard video integrity and content verification, sev-
Deep learning was used to choose and select the accident videos, and eral parts of the video surveillance system can be utilized. Authors
share only necessary information. To improve accuracy, audio and in Kerr et al. (2018) used a combination of blockchain and a digital
image data training was performed together. This not only improved watermarking application to secure video evidence. A prototype camera
the integrity of the video but also integrity of the accident report. was used where it takes part in the creation of blockchain in real
time. End to end protection of video evidence was achieved. Digital
6.2. Video cataloguing watermarks were embedded on to the video prior to transmission.
A communication system was introduced between the devices and
Preserving history through video archives is one of the applications the blockchain nodes which provides a reliable queuing system. This
that blockchain can be used for its uneditable and time stamped feature. allows a scalable and adaptive infrastructure for this system. A video
This data not only catalogued and verified but also secured with hashes management system coupled with blockchain enables proper content
on the blockchain. Striking example of one of its use is in Bui et al. verification with the digitally water marked images.
where an international record of historical videos were preserved using
blockchain through collaboration between several countries. 6.4. Discussion of video integrity applications
Bui et al. in Bui et al. (2019) designed ARCHANGEL for preserving
integrity of digital video fusing computer vision and blockchain. A deep Video integrity applications require a tamper proof and time
network architecture was implemented using a hierarchical attention stamped record enabled by blockchain and specifically automated
auto encoder for computing the temporal content hashes (TCH) for unedited smart contracts. Consensus mechanism were used to verify
the video content. The TCH’s are secured using a proof of authority images and further voted out if fake. Video forensics, content verifica-
blockchain. This was distributed across multiple archives. The appli- tion, content filtering, history preservation through cataloguing were
cation was used to safe-guard against modification of content that solutions to preserving video integrity. In terms of public blockchain,
includes frame truncation or corruption when format shifting occurs. transparency was enabled as all the nodes participating had a copy
To ensure future accessibility, format shift was done on the videos. of the ledger. Private blockchain enabled a private network where
Therefore the original video was transcoded and then compressed using only specified participants were allowed through a core validator.
ffmpeg. The processing happens in such a way that given an input Commonly used blockchain was permissioned blockchain where the
video, its TCH generates hashing in each block, from audio and visual participant need to be verified to take part in the transaction.
content. These hashes were then stored on the blockchain which is Table 3, tabulates the applications in which video integrity preser-
immutable and tamper proof. vation is one of the main gains. As noted from the tables, Video content
Integrity verification was done using the TCHs, SHA-256 hashes processing mostly dealt with the hashing the video and then storing it
and threshold values set, which are stored in a SIP, then stored in a as a transaction in blockchain. Watermarking was the procedure used in
blockchain as a smart contract transaction. The smart contract manages state of the art to identify cameras and verify the content. Blockchain
the access of the data. User access was validated through smart contract was modified in some instances to cater to the storage requirements
12
Table 3
Video/image sharing with blockchain.
Method Application Technique Blockchain Smart Main Limitations
type contract storage
Kumar et al. (2021) Image sharing RCNN training with CT images Ethereum Yes IPFS Cost of transaction, limited
capacity, Non Upgradeable
smart contracts
Zerka et al. (2020) Privacy Encryption of the trained weights Ethereum Yes Cloud archive Cost of transaction,
using Radiomics dataset centralization due to cloud,
interoperability between
institutions, Non
Upgradeable smart
contracts
Tang et al. (2018) Storage Denying fake medical image share Ethereum Yes – Cost of transaction,
interoperability between
institutions, Non
Upgradeable smart
contracts, limited capacity
Shen et al. (2019) Image retrieval Extraction of medical image – Yes On-chain Scalability-transaction time
features and encrypted data increases with number of
storage images, Non Upgradeable
smart contracts
Mehta et al. (2019) Copyright protection Stock photos, P2P sharing Ethereum Yes IPFS Time complexity increases
with number of images,
cost, Non Upgradeable
smart contracts
Esmaili and Javidan (2020) Copyright protection Video sharing systems Ethereum Yes Swarm Time complexity increases
with number of images,
Storage costs, Non
Upgradeable smart
contracts, infrastructure
requirement
Xu et al. (2019) Data sharing with Micro-services Ethereum Yes Database Incomplete
access control decentralization,
Complexity, performance
overhead in terms of
latency, edge and fog
computing challenges
and security demands to safeguard integrity. The most common type where data can be accessed by third party and be vulnerable to several
of blockchain used here was Ethereum and some efficiently used smart attacks. This content may also be prone to copyright violations and can
contracts and chain codes to restrict access or automate the verification be manipulated (Chen et al., 2017). Solution to this was distributed
process. The flaw in most of the applications is that they require high storage through blockchain.
computational resources and are expensive in terms of monetary costs Decentralized storage being a solution for storage, Esmaili et al.
based on the type of blockchain used, as each smart contract execution in Esmaili and Javidan (2020) used swarm, an ethereum blockchain
is considered a transaction. There are multiple steps involved including, supported distributed storage platform that produced no downtime, is
transaction validation, authentication and authorization with consensus secure, non tampered and resistant to censorship. Video access was
execution if any, adding to that the computer vision tasks that are controlled by the smart contract which provides the validity, trans-
performed, eventually having high latency. For a real time process, parency, immutability and integrity of the transactions. A merkle tree
there is need for a study on light weight, low latency techniques. Wa- of video hashes were also stored on the blockchain for immutability and
termarking, hashing and storage in a third party storage, adds on to the verification. Li et al. in Li et al. (2020) used image information from
processing time. Nevertheless, it has been observed that the majority of wireless sensors and converted them to intelligent blocks which were
applications are effective in ensuring data integrity. However, the issue encrypted. The intelligent encrypted algorithms were used to securely
of data reliability and availability remains a concern in this context transmit the images where signature verification was done and then
With latency, real time data may not be available at required speed stored. Medical image data sharing has been one of the most relevant
thereby time, however integrity of the data received even though late application that requires blockchain protection. The following subsec-
can be reliable due to blockchain protection. There is a requirement to tion reviews applications of blockchain for medical image sharing and
boost the efficiency in terms of reliability and availability in achieving further advances in deep learning with medical images.
an optimum model for video integrity.
7.1. Medical image sharing
7. Video/image sharing
Medical data and medical images have been at a rise due to the
Content creation through videos and images have seen a boom increase in internet of medical devices facilitating better health care
in need due to the different video and image sharing platforms like through proper diagnosis and enhanced treatment (Seo and Cho, 2020).
Youtube, Netflix, Vimeo etc, being most popular means of entertain- These medical images are often stored on a centralized network which
ment, news and even education (Esmaili and Javidan, 2020). Video makes it prone to privacy being compromised or even data being
content proliferation is existent at the current times. Multiple open poisoned leading to faulty diagnosis and improper treatment (Guo
platforms are available to upload video content. Open platforms receive et al., 2021). Blockchain based decentralized data with distributed
low profit, and consumers are at risk in terms of privacy. Centralized storage has become popular in recent researches for secure medical
storage is one of the factors that effect the privacy of the consumers image storage and sharing.
13
Fig. 8. Typical video streaming process.
Seo et al. in Seo and Cho (2020) modelled a hyperledger fabric 7.2. Video streaming
architecture a permissioned blockchain for sharing medical images
which are retrieved from a hashed and stored picture archive and Entertainment, music, journalism and currently education, have
communication system (PACS). Image sharing was invoked through exponentiated its need for video streaming applications. The Fig. 8
smart contract. from Ghat (2017) shows how a video streaming process occurs in a
Image retrieval from IoT systems was safe-guarded through use of traditional manner. There is a video source that records the information
blockchain after image was encrypted and stored as transaction in the that is encoded and then saved in a storage location. This is then cached
blocks, which can be retrieved using a retrieval request using smart onto to a content delivery network based on the video content needs,
contract (Shen et al., 2019). Ethereum based medical record storage later to be accessed by the end users video player. A video data may
was one of the applications utilized in Tang et al. (2018) where smart contain a video codec and an audio codec, accompanied by meta data of
contracts were used to deny fake records ensuring security of medical the video with its caption; encoded in several formats and transcoding
images. for this is made to be suitable for end users’ video player. This may
Due to the high need of large scale medical image data for accurate
charge high costs and consume high computational power. This process
AI modelling such as in Abdel Hameed et al. (2022), which is strictly
is usually done by a single central storage. Thus there is an increasing
protected under the GDPR privacy laws, distributed learning or fed-
need for decentralized storage and due to the ever increasing need for
erated learning is implemented (Predd et al., 2006). With distributed
video streaming applications plummeting the online traffic blockchain
learning, images can be shared through an encryption or stored in
is a solution.
a distributed framework as blockchain. Blockchain not only provides
Peer to peer content distribution and blockchain based payment
security as it cannot be tampered with but also privacy is implemented
system have been implemented in several startups. A distributed system
on a private or consortium blockchain.
was introduced in Ghat (2017) that improves the cost and decreases
7.1.1. Distributed learning the computational overhead on a single entity by having a distributed
In distributed learning the data is not shared but the weights of the system of storage. A distributed encoding was also applied where large
process is shared between institutions, this not only secures the data scale video processing can be overcome by re-using unused storage
through distributed learning but also distributed computational costs space at data centres. Problem of high computational complexity and
among institutions involved (Predd et al., 2006). high cost of the current encoders have motivated the vidoecoin Ghat
Zerka et al. in Zerka et al. (2020) presented a scheme for secure (2017), a distributed video processing platform that splits the video
medical data sharing where radiomics dataset images were trained into encoding tasks and to convert them to subtasks and parallelly
and their weights were encyrpted using AES encryption and with a processing them. Open source media framework (ffmpeg) (Cheng et al.,
validated smart contract token, was saved to a archive model cloud, 2012) were run on a secure container, which was safe and cannot
This was implemented using ethereum blockchain. damage host computer. Miners were paid by the clients based on the
Kumar et al. in Kumar et al. (2021) used IPFS system for sending processing completion (Ghat, 2017). A video storage network is also
the local weights and stored the IPFS hashes on blockchain. Smart con- adopted where the storage miners form a storage area network using
tracts were used for distributing the weights. CT images from different their disk space. Distribution miners and storage miners both take part
hospitals were trained using RCNN, where (ROI) region of interest was in delivering the video with lesser computation and cost which can be
identified and trained in a distributed manner. This enabled a privacy negotiated between clients and miners as the videos can be split into
preserving artificial intelligence(AI) with blockchain. payable segments (Ghat, 2017).
14
Table 4
Video integrity preservation applications.
Method Application Technique Blockchain Sim. Smart Modification Limitations
type contract
Kerr et al. (2018) Cataloguing CCTV video Public – No No Scalability, partial
evidence decentralization
through video
watermarking
Xu et al. (2019) Data Micro-services Private Ethereum No No Partial decentralization,
tampering Complexity, performance
security overhead in terms of
latency, edge and fog
computing challenges
Hasan and Salah (2019) Combating Tracing video Public Ethereum No No Decentralized Storage
Deep fake origin Costs, Data privacy,
Complexity of using
multiple technologies
Cremona et al. (2019) Content filtering illegal Public Bitcoin No No Privacy Concerns as filter
verification content may require analysing
content, centralized
control, need to adapt to
dynamic content
Yatskiv et al. (2019) Content Video file – Naive chain No No High network latency,
verification processing, Processing overhead,
Preventing fake security of cloud service
Zelensky et al. (2018) Content Video analysis – Ethereum Yes PROVER app Lack of scalability for large
verification and amount of videos
verification
Danko et al. (2019) Video Video hashing Consortium Fabric Yes No scalability constraints,
forensics Resource intensive, data
transfer security
Lee and Park (2020) Video Efficient – – No Merkle tree Resource intensiveness,
integrity transmission of hash scalability challenges,
data by merkle latency
tree hash
Bui et al. (2019) Cataloguing Hierarchical Public Ethereum No No Resource intensiveness,
and preser- attention interoperability across
vation auto-encoder countries, scalability
for temporal challenges
hashes
Chen et al. (2020a) Combating Data – – – Customized Scalability challenges,
fake verification consensus resource intensiveness,
complex consensus
In another implementation of video streaming, video transcoding Chen et al. (2020b) has four components, the employer, the worker,
was decentralized, where video segments were sent to transcoders the payment scheme and the task allocation mechanism for video
through a smart contract which was programmed to select a transcoder streaming and/or peer to peer video sharing. Employer issues the smart
based on the job and fuel or bitcoin deposited for the transcoding contracts for distributing tasks, the workers receive and complete the
job (Liu et al., 2020b). It not only gives access but decides the val- tasks and are given incentives. This model was tested on a private chain
idators, judges, referees based on delegated Proof of stake (PoS). After for performance in both bidding and task execution performance.
the transcoding process is done, a video streaming and distribution
system verifies through a verification protocol which guarantees that 7.3. Discussion of video/image sharing
transcoding is done correctly. Delivery of videos is performed after
transcoding through mobile edge computing and blockchain Liu et al. Video and/or image sharing require security and privacy, like every
(2020b). other application in computer vision, is one of the prime tasks. Storage
Satish et al. in Sathish et al. (2019) employed microservices, which of the videos were implemented in the blockchain as content hashes
encompassed video stream processing for feature extraction from or an index of the storage on storage systems namely swarm and IPFS.
frames, the enforcement of security policy services, and data access As video content being too large, merkle hashes were used to store as
control with verification. Their objective was to prevent unauthorized transaction, At some point only an event in a video is registered. Image
service requests and tampering by utilizing a permissioned blockchain retrieval, copyright protection while sharing or using stock photos were
network for video sharing. Chen et al. (2020b) and Gu et al. (2018) used some of the applications safeguarded by blockchain through smart con-
blockchain based smart contract for crowd sourcing video transcod- tract implementation. Based on the type of images shared, medical data
ing. This enhanced efficiency thereby leading to lesser computational seemed to have the most significant number of literature where data
resources with distributed computing. In Liu et al. (2020a) a deep privacy was enabled by distributed learning, new learning approach
reinforcement learning (DRL) algorithm was used for selecting the where several organizations do their training with their data and share
transcoder, where videos were converted into multiple formats to pro- the weights for combined inference. Distributed learning was secured
vide fast and reliable service. This was decentralized using blockchain, by blockchain, however data poisoning is one of the possibilities at the
where transcoding process was assigned to bidders. This scheme pro- edge which may make it vulnerable to poor training.
vided quality of service with increased revenue, however, does less in Table 4, describes the applications of video and/or image sharing
terms of saving energy (Liu et al., 2020a). with blockchain. It can be noted that the video sharing systems used
15
Table 5
Overview of common types of blockchain used in the applications.
Blockchain Type Advantage Disadvantage Method
Ethereum Public Transparent Decentralized storage and Has cost called Mudliar et al.
processing, Fault tolerant and has its ether/gas. Has limit for (2018), Iovane et al.
own coupled storage (swarm). scalability, (2018),
Vulnerabilities in smart Delgado-Mohatar
contract, et al. (2019), Tang
et al. (2018), Hasan
and Salah (2019),
Zelensky et al.
(2018) and Bui
et al. (2019)
Ethereum Private Single validator, Uses smart contract, Can risk centralization Xu et al. (2019),
Privacy protected Esmaili and Javidan
(2020) and Zerka
et al. (2020)
Hyperledger Consortium Controlled by a consortium, User access Prone to network Naganuma et al.
Fabric blockchain programmed by chaincode, Has an attacks, Not post (2020) and Sawant
orderer peer and endorser peer, There is quantum secure and Bharadi (2020)
control with decentralization, secure and
private
Bitcoin Public Transparent, Available, Secure, Prone to rewritten Buchmann et al.
Decentralized storage and processing chain and is costly (2017) and Danko
et al. (2019)
ethereum for their transaction. Smart contract was efficiently used to 2021; Farr et al., 2020; Qi et al., 2020,?). Most of the applications
automate the process as well verify and give access for retrieval or involved feature extraction stage in terms of biometrics. Multi-modal
verify a image or video data. Due to the low capacity of storage on- biometrics, that uses multiple biometrics as identity enhanced the
chain, Off-chain storage was widely used. This is a safe alternative to security. Further enhancement was put forth with biometric and nu-
centralized storage. However with the number of steps involved in this merical fusion approach for cryptography which was then used as a
process, there is delay involved increasing latency. Cloud storage was digital signature (Sawant and Bharadi, 2020; Naganuma et al., 2020).
used to store the weights, although encrypted, is vulnerable to security Image encryption proved to not only secure identities but also secure
threats in cloud. blockchain in some applications (Naganuma et al., 2020).
Video streaming applications with blockchain improved the With a unique representation of a person, several literature’s fo-
transcoding process and decreased computational cost with the use cused on using biometrics with blockchain as an identity for govern-
of miners who compete to provide storage space as well as perform mental activities (Buchmann et al., 2017). Video integrity overlapped
transcoding. Video processing was done in a distributed manner by its application in all fields of computer vision. Combating fake seemed
miners. Through deep learning methods with blockchain, quality of to be the prime focus for many applications where content verification,
service, security and privacy was guaranteed for a trustable video watermarking cameras, tracing origin of videos were some of the ways
streaming service. Even though, the cost and computing power is to prevent fake. Hashing video content was noteworthy for its use
distributed, it can lead to delay in video streaming process. Quality of in content verification and video forensics. Video integrity can be
service (QoS) can be a determining factor in terms of choice of video preserved within many applications during surveillance, video sharing,
streaming service along with quality of experience (QoE). Is there a evidence gathering, and video processing. Securing medical images and
possibility of a model that is optimized for the performance, in terms distributed learning for medical data sharing was point of focus in
of QoS and QoE fulfilled fairly is a task for future research. most applications. CT images and X-ray images were shared through
blockchain transactions (Seo and Cho, 2020) Applications in video
8. General discussion streaming, such as distributed video transcoding, not only reduced the
computing cost of the process, but also enabled distributed storage
Data security, integrity and even processing can be safeguarded enhancing the sustainability of storage through delegating tasks to
and accounted for through blockchain systems (Kirillova et al., 2020; peers and renting only the required storage (Sathish et al., 2019; Chen
Shrier et al., 2016; Zikratov et al., 2017). Computer vision is one such et al., 2017; Cheng et al., 2012).
application that can leverage the immutability, both transparency and Storage of big data is a challenge in blockchain as public
private nature of different types of blockchain. Several applications blockchains that have limitations in terms of block size with some
requiring public as well as private network can be leveraged to be used limiting to 1MB which is too low for video content. Several methods
in applications such as securing image data, sharing images as well as like video hashing using merkle hash and other hashing algorithms
video processing information among many other. Tables 1–4 lists out were utilized, however selectivity of content was to be performed.
the applications of computer vision with blockchain available in current Table 6 summarizes the different off-chain storage solutions available
literature. On the type of blockchain used, there is a huge popularity with blockchain enabling a decentralized architecture in balint et al. for
for use in Ethereum blockchain as a public blockchain where anyone storage. Off-chain storage was found to be optimum for video storage
valid can join the network, as well as a private blockchain where access where high resolution image and video files were to be stored. A
control can be restricted (Lee et al., 2020). Hyperledger fabric is used separate physical drive may contain the whole data which is leased
for exercising access control at situations where constrained access from miners. The hash data is only available in the blockchain with a
need to be given for certain nodes (Shalaby et al., 2020). Table 5, lists link to the physical storage location (Hepp et al., 2018; Bálint, 2020;
out in detail the performance of Hyperledger fabric and Ethereum. Esmaili and Javidan, 2020; Bálint, 2020).
In terms of the applications used in computer vision, the dominance Some of the computer vision applications in deployment require
of machine learning and deep learning approaches for image processing a middleware to deal with blockchain network as well as an user
is noted. CNN are widely used in image processing part (Kumar et al., interface to retrieve information. Several architectures with computing
16
Table 6
Off-chain storage for large video footage.
Name Description Advantage Disadvantage Method
IPFS All computer systems are Mutual trust is not Failure to provide data Toutara and
connected by a common required to protect users protection, that is Spathoulas
file system. Contents are data strong enough (2020), Bálint
stored worldwide. Index of (2020), Kumar
content is transaction data. et al. (2021) and
Mehta et al.
(2019)
FileCon Based on IPFS protocol, Secure storage with High cost of Bálint (2020)
storage miners lease out endpoint encryption.pass transaction.
storage. Retrieval miners clients data to the storage
miners. Miners are
renumerated with crypto
coins.
Sia Similar to FileCon, it is a Secure storage with Computationally Bálint (2020)
cloud based endpoint storage complex Reed–Solomon
encryption. Redundant error-correcting scheme.
sharing.
Swarm Peer to peer storage It has zero downtime and Coupled with only Esmaili and
service. A torrent like DDoS resistant. Ethereum. Javidan (2020)
service, incentive drive Appropriate for IoT
and coupled with networks. Its fault tolerant
Ethereum network. and censorship resistant.
Storj Slicing of data to segments Data is secured by private Cost of 1GB is 0.01 Bálint (2020)
and separated into stripes keys. Can store video dollars/month which is
which are grouped into footage. comparatively high.
new erasure codes then
uploaded. Uses asymmetric
encryption.
RIFT Larger block size creation Block size of 5Gb is Not widely Bálint (2020)
achieved through testing available. implemented.
only.
Fig. 9. Medical sharing between organization .
at the edge and fog and at the cloud is implied in several papers. only be shared based on a consensus and transactions created, stored
Fig. 9 illustrates a general architecture of medical image sharing with as encrypted data on the block (on-chain) or on the cloud (Shen et al.,
blockchain that can be deployed at the edge level. The organizations
form a consortium blockchain where the identity or access to infor- 2019). The image can be retrieved through smart contract as in Shen
mation is validated by smart contract. Video and/or image data can et al..
17
Recent developments in hardware and embedded computing has

enabled possibility of embedded blockchain where the device at the
edge can be used as a blockchain node along with performing computer
vision based computations. Some of the devices that are equipped for
general purpose computations are as seen in the Fig. 11.
Raspberry Pi is most common tool used for this purpose but has
restrictions on capacity of storage and GPU computing (Xu et al., 2018).
NVIDIA corporation has released a series of embedded boards named
JETSON nano, Xavier that can be used for heavy GPU computation en-
abling deployment of more accurate models which are trained on larger
datasets for tasks like object detection, classification, abnormal detec-
tion, object recognition and so on. etc (Basulto-Lantsova et al., 2020).
Current developments in light weight models called TinyML (Banbury
et al., 2020) has further enhanced the computational speed and re-
duced complexity of the computations on the edge (Ramachandran and
Krishnamachari, 2018).
Blockchain layer: Depending on the type of blockchain and the
Fig. 10. Layered architecture of blockchain with computer vision applications with network architecture it is integrated with, the blockchain can be im-
common and potential components. plemented in various formats. One approach is to have the blockchain
running on nodes as part of the edge network, while another method
is to deploy it within the fog network. If the blockchain is part of
Storing Video metadata is the most common method of protecting a surveillance system controlled by a regional control station, the
video integrity. Metadata is an important information stored in an EXIF blockchain nodes is several control stations. In the case of a national
(Exchangeable Image File) format. The metadata has information on identity containing blockchain, it would be the public institutions.
device captured, the settings of the video captured like shutter speed Smart contracts are used to control access to the blockchain in several
and focal length, the history of capture that includes the time and levels.
location of the capture. Apart from this information that are similar Computer vision layer: The computer vision layer is where the
to images, they have container formats like AVI and MP4, that hold image processing, pattern recognition, object detection is performed.
the codec information, video, and audio streams. This information does Deep learning and machine learning models are deployed in this layer.
not require high storage capacity but is very crucial to identification of The images captured are inferred for object recognition , detection,
authenticity, origin and formatting of the image or video. This when and identification or matching based on the application use case. Core
added with blockchain security by storing the metadata on the block information is extracted and encrypted. With the use of an application
can enhance the provenance of videos. interface (API) and decentralized applications (dAPPs) they are stored
on to the blockchain or a decentralized storage later to be hashed on
8.1. Blockchain-computer-vision implementation layers to the blockchain Paralkar et al. (2018).
Application interface layer: The application interface acts as the
It can be evident that a generalized layered architecture can be link between the blockchain and the computer vision application.
charted from all the models. In Ramachandran and Krishnamachari Decentralized applications are created where blockchain API’s and
(2018), a generalized blockchain for IoT is layered with a server and Computer vision application API’s are visualized. Smart contracts are
back-end layers, edge device and the end device. The server and back- deployed on this layer and communicate with the blockchain using an
end constitutes the part of the application layer that performs the oracle or a middle ware. Applications like Infura, Ganache are used
application layer functions. Edge computing is performed in a separate for interfacing with blockchain through private blockchain simulations
layer with network connections in another layer. With this in perspec- and applications (Hu et al., 2018; Zupan et al., 2017; Taş and Tan-
tive and summarizing the application uses cases, Fig. 10 is an overview rıöver, 2019). Scripting languages like solidity and Golang are used
of layers of blockchain with computer vision applications and common to program smart contract and dApps. The decentralized storage like
tools used in each layer of network. The layers are divided into four IPFS and swarm use API’s to interface with the blockchain and other
layers, Image/video capturing layers, computer vision layer, blockchain applications. Using Application interface, a micro-service architecture
layer and application interface. The following describes each layer in can also be created where different tasks can be done at different
brief. nodes, performing varied tasks as in Nagothu et al. (2018). A typical
Image/video capturing layer: There can be several image captur- usage of blockchain with access control using smart contracts for
ing devices like of that PTZ cameras, CCTV cameras, mobile phone person re-identification as illustrated in Fig. 12. Other forms of layered
cameras, dash cams and UAV’s/drones equipped with cameras all architecture for blockchain based application can be moulded based on
forming part of the blockchain network. This layer is at the core of use case. Two layered federated learning architecture with hyperledger
the blockchain where data is captured. There can be three ways this fabric blockchain for security can be seen in Feng et al. (2021). MNIST
data is used for the applications. data was trained at several nodes and weights shared through secure
hyperledger fabric blockchain.
• The captured data is sent to the computer vision layer where large
scale image/video processing is performed. Machine an d deep 8.2. Evaluation metrics
learning inference can be done at this layer where the pretrained
models are stored for inference. As with every application, the success or usability depends on its
• The captured data is stored in a decentralized storage where the efficiency and effectiveness to perform task assigned. Performance of
index is hashed on to the blockchain. blockchain is commonly measured based on the success rate, aver-
• The captured devices can also be embedded with image/video age latency, throughput, cost, and resource consumption (Dabbagh
processing capability that can then use an oracle or middle ware et al., 2020; Pongnumkul et al., 2017). Image and/or Video process-
to store the data/metadata as transactions on the blockchain ing part efficiency depends on the type of task it performs. Tasks
Mendki (2019). like object detection and image segmentation have metrics related
18
Fig. 11. Commonly used experimental edge devices for camera embedded computing.
Fig. 12. Person Re-identification using smart contracts for image retrieval and storage to provide access control.
to the process it performs like accuracy of detection measured by averaged to quantify the resource consumed during the blockchain
mean average precision, number of frames processed per second etc. process. Storage capacity and overall complexity in computation both
The following are common evaluation metrics that can be used in time and memory are measured for this purpose (Dorri et al., 2017).
blockchain applications with computer vision: Scalability: The number of nodes that be part of a blockchain is
Success rate: The Success rate is typically measured in the essential in designing an efficient blockchain application. Most private
blockchain layer as the number of successful transaction performed out blockchains can control the number of nodes in the network thus main-
of 100 transactions (Dabbagh et al., 2020). taining the smaller chain and improving latency. Public blockchains on
Average latency: The time taken to execute the code after initial-
the other hand can be unlimited, however each blockchain model has
ization is usually measured as latency. With smart contracts involved
its own set scale. Multi-chain approaches are used in some applications
and an added layer of image processing, the latency is found to usually
for enhancing scalability (Badr et al., 2019).
increase in applications. Latency can be predicted using architecture
Security metrics: Applications like video integrity, data sharing
modelling and using testnets and simulations for which framework was
proposed in Yasaweerasinghelage et al. (2017). and biometrics are most vulnerable to security threats although mit-
Throughput: The measure of number of transactions performed per igated in terms of architecture by blockchain. There are loopholes in
second is termed as throughput. Several blockchains have a predefined the system that need to be addressed. Quantifying how secure is the
number of transactions set. The addition of another layer of compu- blockchain with computer vision model requires security be dealt with
tation in terms of smart contract, verification and inference on the in all the phases of the process starting form where AI is applied, to
computer vision layer, this may vary (Gupta et al., 2020). identifying or capturing the required data to be stored or indexed.
Resource Consumption: Resource consumption is a key factor in The type of blockchain influences the blockchain model used and
usability of the system. The memory used and the CPU usage are so the security implications too. Abdelwahed et al. in Abdelwahed
19
et al. (2020) provides methodology for measuring cyber-security risks sensitive information and enable malicious activity. The existence of
involved in blockchain applications. Security in micro-services architec- private keys in blockchain serves as one of the defence mechanisms
ture were divided as mining services and security policy services (Shen in the blockchain architecture. Signed transactions ensure validity of
et al., 2019). Access control in blockchain provides authorization, the information. Nodes performing mining mechanism can reject a
the cryptographic keys provide authentication, the immutability pro- transaction of an unauthorized entity (Li et al., 2020). Supplementary
vides reliability and consensus mechanism provides validation and measures on the blockchain network such as timestamps, nonces and
verification. unique chain identification such as the chainID incorporated onto the
In terms of the mining services present in blockchain used for val- ethereum network mitigate any replay attack ensuring integrity of
idation and verification provides security for the private permissioned the data secured (Anita. and Vijayalakshmi., 2019; Chalaemwongwan
blockchain from unauthorized miners and transaction. The security and Kurutach, 2018). Timestamp, nonce and chainID are included in
policies are defined by micro-services each orchestrating policies that the transaction data of the blockchain. Every new transaction on the
involve authentication and access control. This was advantageous in en- blockchain undergoes scrutiny to verify the uniqueness of the nonce, a
hancing scalability and heterogeneity in surveillance systems through unique number, and the timestamp, which records the time the block
light-weight security solutions that are flexible and inter-operable. was created. These values are embedded in the header of the block
Performance evaluation of most of the models in the applications within the blockchain. If the uniqueness of the nonce and timestamp is
surveyed were based on security, throughput, latency and scalability. not valid, the transaction is rejected, thus preventing a replay attack.
Each layer of the blockchain with computer vision model had its own Impersonation and Sybil Attacks: Impersonation and Sybil attacks
evaluation metric. Al-Sahan et al. (2020) measured the latency for the can be thwarted on an Ethereum framework as the Ethereum holds
whole model, where as (Bálint, 2020) compared the security of off- a decentralized Ethereum name system (ENS) which contains all the
chain storage with cost involved. Esmaili et al. performed security, scal- public keys generated. This with the use of biometric keys can enhance
ability and time complexity analysis as performance metrics (Esmaili the security of Ethereum.
and Javidan, 2020). Privacy: With consortium blockchains, valid parties can be given
access to images or videos based on their authority level as seen in
8.3. Advantages of blockchain for computer vision many applications with biometrics and video surveillance. Restricting
access to unauthorized parties safeguards privacy of the content.
Computer vision applications with blockchain enhance the applica- Decentralized storage: Huge amounts of data retrieved from Video
tion in several ways. The striking features of blockchain that enable this footage and video streaming need not be stored at one infrastructure.
enhancement are the following: Several secure non-cloud, decentralized storage systems are available
Integrity: Video evidence is kept and can be traced back to its to store the videos.
origin, all transaction are verified and hashed enabling trust in video Decentralized processing: Secure sharing of gradient of machine
content in the blockchain. Smart contracts are used to control ac- learning models enables federated or distributed learning which pro-
cess, verify content and automate the process which in itself is uned- vides opportunities for secure big data analysis and collaborative train-
itable after deployment even by the author themselves (Christidis and ing producing much more accurate results as seen in distributed learn-
Devetsikiotis, 2016). ing for medical images.
Accountability: Every video catalogued or stored can be traced Fault tolerance: Duplication of records enables fault tolerance in
back to its origin as blockchain is an uneditable ledger which is time an decentralized network (Esmaili and Javidan, 2020).
stamped, which makes the blockchain network accountable for the Transparency and Trust: Visibility of all the transactions in each
transactions. node provides trust to the participants despite being decentralized (Es-
Non-repudiation: As the ledger is uneditable and cryptographically maili and Javidan, 2020).
signed by the author as private key as well as public key there is no
contest in the origin of videos. Biometric keys provide an even more 8.4. Challenges and future directions
secure key where the author can be identified.
Authorization: Smart contracts can be programmed to perform Blockchain has brought forth many advantages to computer vision
authorization as well function with authorization, where a certain as stated above, however every application has its own challenges and
authority can only execute the functions in a contract. This enables ac- limitations as stated in columns of Tables 1–4.
countability as well as access control in some situations where privacy Real time analysis: Most state of art applications of computer vi-
and security of the video information need to be taken care of as in sion with blockchain involve architectures that need to be implemented
video integrity applications as well as video surveillance. in real time for further analysis. Although decentralized architectures,
Availability: Public blockchain network is transparent, all nodes some of the applications are prone to centralization due to the use of a
have access to the ledger and provenance data is available for the central node as validator such as in private blockchains.
participants. Denial of service(DoS) and Distributed denial of ser- Scalability: Scalability being its first challenge, blockchain architec-
vice(DDoS) are prevented in this network due to its global placement. ture’s capacity for transaction processing is very low as of now along
As there is no single point of failure, there is availability of data all the with its block size amounting to 1Mb like that of bitcoin Wu et al.
time. (2020). With most of its applications in identity management, there
Replay Attacks: is a need for high scalability of the blockchain architecture. Public
Replay attacks occur when a communication is captured and played blockchains like Ethereum and Bitcoin was calculated to be 200 GB as
at a later time (Smith et al., 2015). In the case of blockchain, transac- its scalable limit (Delgado-Mohatar et al., 2019). As with the popularity
tion data can be subjected to replay attacks after a hard fork occurs. of many blockchains, the size of blockchains have been growing which
Two distinct blockchain branch out for the original chain. In this case can be a bottle neck for several applications (Delgado-Mohatar et al.,
replay attack materializes when malicious nodes intercept a valid data 2019). To over come scalability limitation of the blockchain, parallel
transmission on one chain and replicate and re-transmit at a later junc- processing was a solution put forth in Akbari et al. (2017), which can
ture . In the case of computer vision applications, this could be harmful be used for several applications such as video transcoding.
when data such as biometric templates, image hashes are replayed by Unexpected Security risks: Security with biometric keys can be a
malicious nodes compromising integrity resulting in erroneous data added fortress to blockchain, however there is a possibility of attack
and falsified information. It is imperative that replay attacks should that is not studied yet. This being an individuals key can make them
be recognized and thwarted as they provide unauthorized access to vulnerable to frauds and thefts. The type of blockchain may determine
20
the cost involved in the process and some smart contracts when ex- face. This leads to ageing of biometric templates which can cause high
ecuted need to be rewarded accumulating cost (Albert et al., 2020). error rate in recognition and consequently effect efficient access control
Implementation of a single national identity with biometric data and and identity management (Buchmann et al., 2017). Losing the key to
breeder documents. Further enhancing security through quantum proof blockchain can cause considerable data loss.
blockchain like IOTA as in national identify breeder documents (Buch- Post quantum Security: Post quantum security is a big concern for
mann et al., 2017). Furthering research for more safer multi modal the unexpected security risks that can arise for blockchain applications
biometrics for encryption keys in blockchain and so on (Iovane et al., as well as computer vision. Current applications are not quantum se-
2018). Biometric key can merge for copyright protection in video cure. Buchmann et al. in Buchmann et al. (2017) suggested a hard fork
streaming and video transcoding through water marking the segments on blockchain for applying a post quantum resistant digital signature
of the videos for safety of the video from tampering in the process of in the metadata that is the integrated biometric embedded breeder
transcoding (Kerr et al., 2018). document. An example of post quantum resistant digital signature as
Immutability of smart contracts: Smart contracts being imple- stated in Buchmann et al. (2017) would be XMSS and SPHINCS. XMSS
mented in almost all applications of computer vision for automation, (eXtended Merkle Signature Scheme) where private key changes with
verification or access control, need to be checked carefully to avoid every signature generation (Hülsing et al., 2018). SPHINCS is a stateless
potential bugs (Mohanta et al., 2018). Frauds and attacks are malicious long term hash-based signature scheme which is resistant to quantum
behaviours that should be taken into consideration when smart contract threats (Bernstein et al., 2015). Blockchain architectures viable for IoT
developers need to be aware of the contract’s interaction patterns to like IOTA which is quantum secure can be used for Computer vision
mitigate potential loses due the malicious behaviours (Destefanis et al., applications deployed at the edge.
2018). It is impossible to make revisions once deployed (Mohanta With all these taken into consideration, the processing capability
et al., 2018). Careful planning needs to be done in order to have of the blockchain depends on type of blockchain used. Ethereum for
secure functioning blockchain system. Time and assets are also involved example, can run only 12 transactions per second which when com-
in keeping up contracts along with system expenses of the whole pared to large amount of real time video footage generated is not
framework (Michelin et al., 2020). sufficient. And so the type of blockchain used should cater to the ap-
Computational Cost: Computer vision applications require high plication requirements (Delgado-Mohatar et al., 2019). Specific to the
performance hardware and high computational cost in most applica- applications involved, for an image retrieval system, encrypted image
tions (Afif et al., 2020). This causes an overhead in processing along features for data modelling was suggested as a future work in Shen et al.
with the latency involved in the computer vision with blockchain ap- this is predicted to preserve privacy and security further. Rewarding
plications where transactions are exchanged, validated and confirmed members based on contribution based on shapely method was a future
with an consensus algorithm along with automated smart contracts and direction (Shen et al., 2019) to implement. Image processing systems in
in some architectures API’s which bridge between the different layers of computer applications can be further enhanced for specificity as noted
the architecture. Enhancing security can be cause of high latency and in Al-Sahan et al. (2020). Image frauds have been dealt with in many
thereby delay in output (Gupta et al., 2021). Although, some image applications using hashing. However, image transformation was found
processing applications implemented have had minimal delay as in Al- to be an issue for future research (Mehta et al., 2019).
Sahan et al. (2020). Removing one property of blockchain like mining Distributed learning is an application that is blooming in the cur-
where it is not necessary can be a point forward for future applications. rent. However there is trust involved in the image data and the weights
Proactive Predictive surveillance: In terms of surveillance sys- shared among two entities at the edge which leads to need for better
tems, a proactive predictive surveillance is a future work described in encryption methodologies for securing the data and safeguarding which
the state of art with blockchain protection. Although light weight mod- can be done by enhancing blockchain Kairouz et al. (2019).
els are utilized for blockchain in video surveillance in terms of security, Light weight accurate mechanisms are the need for now due to
feature sharing can be a direction in terms of person re-identification the lack of storage and security issues of off-chain storage, where
and gait recognition tasks. Secure video summarization (Elharrouss on-chain storage need to be used (Singh et al., 2020). In addition,
et al., 2019) for suspect capturing from multiple cameras and pri- latency involved in large systems need to be mitigated with use of
vacy preservation through this approach with blockchain can enhance several front-end and back-end applications adding to the delay time.
suspicious activity tracking and immediate response to critical inci- A comprehensive robust system without trade off in terms of latency,
dents. Biometric encryption of images and videos for content preser- throughput and computational overhead, accuracy of output should be
vation and video integrity preservation can be used with blockchain the aim for computer vision applications.
for authentication of video content.
Need for Standardization: Once the application is deployed, there 9. Conclusion
is a need for privacy, security and storage, and evidential data storing
procedures to be standardized for law enforcement through watermark- This paper, provides a overview of blockchain applications with
ing camera vendors, using cameras with embedded chips for creating computer vision and potential future works. It discriminates the appli-
on-camera blocks etc (Kerr et al., 2018). There is a need for further re- cations based on the use cases and then elaborates each application.
search in applications related to video sharing in sensitive applications An overview and generalized model was constructed for each appli-
like educational videos, videos that contain protected content, videos cation with that there was a deep delve into performance metrics,
that have minors involved etc. layered architecture, and the pros and cons. Compared to a centralized
Storage: Cloud storage is often used in this systems due to large architecture a trade-off is eminent in terms of security, throughput
amount of storage required for video content (Zerka et al., 2020). and scalability as well as latency based on the application scenario.
Several peer to peer centralized storage are analysed for this that can Several generalized architectures are presented in this paper as well
be leveraged for future applications (Bálint, 2020), Table 5 describes as some future directions to mitigate the drawbacks. Discussion in
the types of storage applications available that are decentralized and each section summarizes, and presents the challenges that need to
commonly used with blockchain. Storage can be costly, insecure and be overcome to optimize the blockchain performance with computer
some computationally complex codes for error checking which needs vision applications. In conclusion there is more room for research in
to be dealt with as future work. each category reviewed and there is need for the use of this revolu-
Biometrics template ageing: Biometric being an unique identity tionary technology, blockchain, to merge with computer vision into
of a person is applied for authentication as well as digital signature; more research paths. Several applications can further be used with
it should be noted that certain features are effected by ageing such as the introduction of distributed or federated blockchain and encryption
21
techniques with fusion biometrics to secure the training of massive Badr, A., Rafferty, L., Mahmoud, Q.H., Elgazzar, K., Hung, P.C.K., 2019. A permissioned
amount of data captured by different organizations. blockchain-based system for verification of academic records. In: 2019 10th IFIP
International Conference on New Technologies, Mobility and Security. NTMS, pp.
Future direction: In the context of establishing a secure and private
1–5. http://dx.doi.org/10.1109/NTMS.2019.8763831.
data transmission and communication system, blockchain emerges as Bálint, K., 2020. Modern, decentralized blockchain-based solutions for saving video
a promising solution for computer vision applications. Nevertheless, footage. In: 2020 IEEE 18th International Symposium on Intelligent Systems and
there exists a crucial need to address scalability issues and accommo- Informatics. SISY, IEEE, pp. 11–14.
Banbury, C.R., Reddi, V.J., Lam, M., Fu, W., Fazel, A., Holleman, J., Huang, X.,
date the storage of large datasets with real time processing a funda- Hurtado, R., Kanter, D., Lokhmotov, A., et al., 2020. Benchmarking tinyml systems:
mental requirement for computer vision tasks. Future research should Challenges and direction. arXiv preprint arXiv:2003.04821.
investigate the trade-off between these requirements and optimizing Basulto-Lantsova, A., Padilla-Medina, J.A., Perez-Pinal, F.J., Barranco-Gutierrez, A.I.,
solutions for a more efficient system. Blockchain technology applica- 2020. Performance comparative of OpenCV template matching method on Jetson
TX2 and Jetson nano developer kits. In: 2020 10th Annual Computing and
tions in computer vision include various domain of those in public
Communication Workshop and Conference. CCWC, IEEE, pp. 0812–0816.
and private systems. A critical area for future exploration is achieving Bernstein, D.J., Hopwood, D., Hülsing, A., Lange, T., Niederhagen, R., Pa-
interoperability between these diverse blockchain networks. Access in pachristodoulou, L., Schneider, M., Schwabe, P., Wilcox-O’Hearn, Z., 2015.
a controlled and timely manner of real time data secured on blockchain SPHINCS: practical stateless hash-based signatures. In: Annual International Con-
ference on the Theory and Applications of Cryptographic Techniques. Springer, pp.
such as surveillance video footage, biometrics identification data, med-
368–397.
ical images and live streaming data. Additionally, analysis of the data Bolle, R.M., Connell, J.H., Pankanti, S., Ratha, N.K., Senior, A.W., 2013. Guide to
and availability of the resultant data in a secure network. Biometrics. Springer Science & Business Media.
As the field evolves, it becomes imperative to develop and up- Buchmann, N., Rathgeb, C., Baier, H., Busch, C., Margraf, M., 2017. Enhancing breeder
document long-term security using blockchain technology. In: 2017 IEEE 41st
date security mechanisms and build advanced cryptographic methods Annual Computer Software and Applications Conference, Vol. 2. COMPSAC, IEEE,
that can withstand potential post-quantum security breaches, ensuring pp. 744–748.
the long-term security of blockchain-based systems. Further research Bui, T., Cooper, D., Collomosse, J., Bell, M., Green, A., Sheridan, J., Higgins, J., Das, A.,
avenues include the creation of decentralized image databases, the inte- Keller, J., Thereaux, O., et al., 2019. Archangel: Tamper-proofing video archives
using temporal content hashes on the blockchain. In: Proceedings of the IEEE/CVF
gration of AI-generated smart contracts, and the seamless incorporation Conference on Computer Vision and Pattern Recognition Workshops.
of image and video analysis capabilities facilitated by blockchain tech- Chalaemwongwan, N., Kurutach, W., 2018. A practical national digital ID framework
nology. These areas represent exciting opportunities for innovation and on blockchain (NIDBC). In: 2018 15th International Conference on Electrical Engi-
advancement in the intersection of blockchain and computer vision. neering/Electronics, Computer, Telecommunications and Information Technology.
ECTI-CON, pp. 497–500. http://dx.doi.org/10.1109/ECTICon.2018.8620003.
Chauhan, A., Malviya, O.P., Verma, M., Mor, T.S., 2018. Blockchain and scalability. In:
Declaration of competing interest 2018 IEEE International Conference on Software Quality, Reliability and Security
Companion. QRS-C, IEEE, pp. 122–128.
Chen, T.Y., Biglari-Abhari, M., Kevin, I., Kai Wang, A., 2017. Trusting the computer
The authors declare that they have no known competing finan- in computer vision: A privacy-affirming framework. In: Proceedings of the IEEE
cial interests or personal relationships that could have appeared to Conference on Computer Vision and Pattern Recognition Workshops. pp. 56–63.
influence the work reported in this paper. Chen, Q., Srivastava, G., Parizi, R.M., Aloqaily, M., Al Ridhawi, I., 2020a. An incentive-
aware blockchain-based solution for internet of fake media things. Inf. Process.
Manage. 57 (6), 102370.
Acknowledgement Chen, Y., Yin, H., Xiang, Y., Ren, W., Ren, Y., Xiong, N.N., 2020b. CVT: A crowdsourc-
ing video transcoding scheme based on blockchain smart contracts. IEEE Access 8,
220672–220681.
This research work was made possible by research grant support Cheng, J.C., Lee, N.Y., Chi, C., Chen, Y.H., 2018. Blockchain and smart contract
(QUEX-CENG-SCDL-19/20-1) from Supreme Committee for Delivery for digital certificate. In: 2018 IEEE International Conference on Applied System
and Legacy (SC) in Qatar. Invention. ICASI, IEEE, pp. 1046–1051.
Cheng, Y., Liu, Q., Zhao, C., Zhu, X., Zhang, G., 2012. Design and implementa-
tion of mediaplayer based on ffmpeg. In: Software Engineering and Knowledge
References Engineering: Theory and Practice. Springer, pp. 867–874.
Chowdhury, M.J.M., Ferdous, M.S., Biswas, K., Chowdhury, N., Kayes, A., Alazab, M.,
Abdel Hameed, M., Hassaballah, M., Hosney, M.E., Alqahtani, A., 2022. An AI-enabled Watters, P., 2019. A comparative analysis of distributed ledger technology
Internet of Things based autism care system for improving cognitive ability of platforms. IEEE Access 7, 167930–167943.
children with autism spectrum disorders. Comput. Intell. Neurosci. 2022. Christidis, K., Devetsikiotis, M., 2016. Blockchains and smart contracts for the Internet
Abdelwahed, I.M., Ramadan, N., Hefny, H.A., 2020. Cybersecurity risks of blockchain of Things. IEEE Access 4, 2292–2303.
technology. Int. J. Comput. Appl. 975, 8887. Cremona, K., Tabone, D., De Raffaele, C., 2019. Cybersecurity and the blockchain:
preventing the insertion of child pornography images. In: 2019 International
Abuhashim, A., Tan, C.C., 2020. Smart contract designs on blockchain applications.
Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery.
In: 2020 IEEE Symposium on Computers and Communications. ISCC, pp. 1–4.
CyberC, IEEE, pp. 197–204.
http://dx.doi.org/10.1109/ISCC50000.2020.9219622.
Dabbagh, M., Kakavand, M., Tahir, M., Amphawan, A., 2020. Performance analysis of
Afif, M., Said, Y., Atri, M., 2020. Computer vision algorithms acceleration using graphic
blockchain platforms: Empirical evaluation of hyperledger fabric and ethereum. In:
processors NVIDIA CUDA. Cluster Comput. 1–13.
2020 IEEE 2nd International Conference on Artificial Intelligence in Engineering
Akbari, E., Wu, Q., Zhao, W., Arabnia, H.R., Yang, M.Q., 2017. From blockchain to
and Technology. IICAIET, pp. 1–6. http://dx.doi.org/10.1109/IICAIET49801.2020.
internet-based voting. In: 2017 International Conference on Computational Science
9257811.
and Computational Intelligence. CSCI, IEEE, pp. 218–221.
Danko, D., Mercan, S., Cebe, M., Akkaya, K., 2019. Assuring the integrity of videos
Al-Sahan, L., Al-Jabiri, F., Abdelsalam, N., Mohamed, A., Elfouly, T., Abdallah, M., from wireless-based IoT devices using blockchain. In: 2019 IEEE 16th International
2020. Public security surveillance system using blockchain technology and ad- Conference on Mobile Ad Hoc and Sensor Systems Workshops. MASSW, IEEE, pp.
vanced image processing techniques. In: 2020 IEEE International Conference on 48–52.
Informatics, IoT, and Enabling Technologies. ICIoT, IEEE, pp. 104–111. Deepak, K., Badiger, A.N., Akshay, J., Awomi, K.A., Deepak, G., Kumar, H., 2020.
Albert, E., Correas, J., Gordillo, P., Román-Díez, G., Rubio, A., 2020. GASOL: gas anal- Blockchain-based management of video surveillance systems: A survey. In: 2020
ysis and optimization for ethereum smart contracts. In: International Conference 6th International Conference on Advanced Computing and Communication Systems.
on Tools and Algorithms for the Construction and Analysis of Systems. Springer, ICACCS, IEEE, pp. 1256–1258.
pp. 118–125. Delgado-Mohatar, O., Fierrez, J., Tolosana, R., Vera-Rodriguez, R., 2019. Biometric
Andreev, R., Andreeva, P., Krotov, L., Krotova, E., 2018. Review of blockchain template storage with blockchain: A first look into cost and performance tradeoffs.
technology: Types of blockchain and their application. Intellekt. Sist. Proizv. 16 In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern
(1), 11–14. Recognition Workshops.
Anita., N., Vijayalakshmi., M., 2019. Blockchain security attack: A brief survey. In: Destefanis, G., Marchesi, M., Ortu, M., Tonelli, R., Bracciali, A., Hierons, R., 2018.
2019 10th International Conference on Computing, Communication and Networking Smart contracts vulnerabilities: a call for blockchain software engineering? In: 2018
Technologies. ICCCNT, pp. 1–6. http://dx.doi.org/10.1109/ICCCNT45670.2019. International Workshop on Blockchain Oriented Software Engineering. IWBOSE,
8944615. IEEE, pp. 19–25.
22
Dhiran, A., Kumar, D., Arora, A., et al., 2020. Video fraud detection using blockchain. Hepp, T., Sharinghousen, M., Ehret, P., Schoenhals, A., Gipp, B., 2018. On-chain vs.
In: 2020 Second International Conference on Inventive Research in Computing off-chain storage for supply-and blockchain integration. it-Inf. Technol. 60 (5–6),
Applications. ICIRCA, IEEE, pp. 102–107. 283–291.
Dorri, A., Kanhere, S.S., Jurdak, R., 2017. Towards an optimized blockchain for IoT. Hu, Y.C., Lee, T.T., Chatzopoulos, D., Hui, P., 2018. Hierarchical interactions between
In: 2017 IEEE/ACM Second International Conference on Internet-of-Things Design ethereum smart contracts across testnets. In: Proceedings of the 1st Workshop on
and Implementation. IoTDI, IEEE, pp. 173–178. Cryptocurrencies and Blockchains for Distributed Systems. pp. 7–12.
Drescher, D., 2020. Blockchain Basics: A Non-Technical Introduction in 25 Steps. Gildan Hughes, L., Dwivedi, Y.K., Misra, S.K., Rana, N.P., Raghavan, V., Akella, V., 2019a.
Audio. Blockchain research, practice and policy: Applications, benefits, limitations, emerg-
Elharrouss, O., Al-Maadeed, N., Al-Maadeed, S., 2019. Video summarization based on ing research themes and research agenda. Int. J. Inf. Manage. 49, 114–129.
motion detection for surveillance systems. In: 2019 15th International Wireless http://dx.doi.org/10.1016/j.ijinfomgt.2019.02.005, URL: http://www.sciencedirect.
Communications & Mobile Computing Conference. IWCMC, IEEE, pp. 366–371. com/science/article/pii/S0268401219302014.
Elharrouss, O., Almaadeed, N., Al-Maadeed, S., 2020. LFR face dataset: Left-front-right Hughes, L., Dwivedi, Y.K., Misra, S.K., Rana, N.P., Raghavan, V., Akella, V.,
dataset for pose-invariant face recognition in the wild. In: 2020 IEEE International 2019b. Blockchain research, practice and policy: Applications, benefits, limitations,
Conference on Informatics, IoT, and Enabling Technologies. ICIoT, IEEE, pp. emerging research themes and research agenda. Int. J. Inf. Manage. 49, 114–129.
124–130. Hülsing, A., Butin, D., Gazdag, S.L., Rijneveld, J., Mohaisen, A., 2018. XMSS: eXtended
Esmaili, M., Javidan, R., 2020. A distributed blockchain-based video sharing system Merkle signature scheme. In: RFC 8391. IRTF.
with copyright, integrity, and immutability. In: 2020 8th Iranian Joint Congress Iovane, G., Bisogni, C., De Maio, L., Nappi, M., 2018. An encryption approach using
on Fuzzy and Intelligent Systems. CFIS, IEEE, pp. 86–92. information fusion techniques involving prime numbers and face biometrics. IEEE
Farr, Z., Azab, M., Samir, E., 2020. Blockchain-based cooperative autonomous detec- Trans. Sustain. Comput. 5 (2), 260–267.
tion of suspicious vehicles. In: 2020 11th IEEE Annual Information Technology, Islam, M.N., Kundu, S., 2018. Preserving IoT privacy in sharing economy via smart
Electronics and Mobile Communication Conference. IEMCON, IEEE, pp. 0188–0192. contract. In: 2018 IEEE/ACM Third International Conference on Internet-of-Things
Feng, X., Jiang, Y., Yang, X., Du, M., Li, X., 2019. Computer vision algorithms and Design and Implementation. IoTDI, IEEE, pp. 296–297.
hardware implementations: A survey. Integration 69, 309–320. Islam, A., Shin, S.Y., 2019. BUAV: A blockchain based secure UAV-assisted data
Feng, L., Yang, Z., Guo, S., Qiu, X., Li, W., Yu, P., 2021. Two-layered blockchain acquisition scheme in Internet of Things. J. Commun. Netw. 21 (5), 491–502.
architecture for federated learning over mobile edge network. IEEE Network 1–14. http://dx.doi.org/10.1109/JCN.2019.000050.
http://dx.doi.org/10.1109/MNET.011.2000339. Jacobovitz, O., 2016. Blockchain for Identity Management. The Lynne and William
Fitwi, A., Chen, Y., 2021. Secure and privacy-preserving stored surveillance video Frankel Center for Computer Science Department of Computer Science. Ben-Gurion
sharing atop permissioned blockchain. arXiv preprint arXiv:2104.05617. University, Beer Sheva.
Fitwi, A., Chen, Y., Zhu, S., 2019. A lightweight blockchain-based privacy protection Jain, A.K., Flynn, P., Ross, A.A., 2007. Handbook of Biometrics. Springer Science &
for smart surveillance at the edge. In: 2019 IEEE International Conference on Business Media.
Blockchain. Blockchain, IEEE, pp. 552–555. Jain, A.K., Nandakumar, K., Nagar, A., 2008. Biometric template security. EURASIP J.
Gallo, P., Pongnumkul, S., Nguyen, U.Q., 2018. BlockSee: Blockchain for IoT video Adv. Signal Process. 2008, 1–17.
surveillance in smart cities. In: 2018 IEEE International Conference on Environment Jain, A.K., Ross, A.A., Nandakumar, K., 2011. Introduction to Biometrics. Springer
and Electrical Engineering and 2018 IEEE Industrial and Commercial Power Science & Business Media.
Systems Europe. EEEIC/I&CPS Europe, IEEE, pp. 1–6. Jeong, Y., Hwang, D., Kim, K.H., 2019. Blockchain-based management of video
García, C.G., Meana-Llorián, D., G-Bustelo, B.C.P., Lovelle, J.M.C., Garcia-Fernandez, N., surveillance systems. In: 2019 International Conference on Information Networking.
2017. Midgar: Detection of people through computer vision in the Internet of ICOIN, IEEE, pp. 465–468.
Things scenarios to improve the security in smart cities, smart towns, and smart Kairouz, P., McMahan, H.B., Avent, B., Bellet, A., Bennis, M., Bhagoji, A.N.,
homes. Future Gener. Comput. Syst. 76, 301–313. Bonawitz, K., Charles, Z., Cormode, G., Cummings, R., et al., 2019. Advances and
Garg, K., Saraswat, P., Bisht, S., Aggarwal, S.K., Kothuri, S.K., Gupta, S., 2019. A com- open problems in federated learning. arXiv preprint arXiv:1912.04977.
paritive analysis on e-voting system using blockchain. In: 2019 4th International Kamble, M.R., Sailor, H.B., Patil, H.A., Li, H., 2020. Advances in anti-spoofing: from
Conference on Internet of Things: Smart Innovation and Usages. IoT-SIU, IEEE, pp. the perspective of ASVspoof challenges. APSIPA Trans. Signal Inf. Process. 9.
1–4. Ke, X., Zhang, Y., 2020. Fine-grained vehicle type detection and recognition based on
Ghat, D., 2017. VideoCoin-a decentralized video encoding, storage, and content dense attention network. Neurocomputing 399, 247–257.
distribution network. Kerr, M., Han, F., van Schyndel, R., 2018. A blockchain implementation for the
Gollapudi, S., 2019. Deep learning for computer vision. In: Learn Computer Vision using cataloguing of cctv video evidence. In: 2018 15th IEEE International Conference
OpenCV. Springer, pp. 51–69. on Advanced Video and Signal Based Surveillance. AVSS, IEEE, pp. 1–6.
Gu, Y., Chen, J., Wu, X., 2018. An implement of smart contract based decentralized Kim, T., Jung, I.Y., Hu, Y.C., 2020. Automatic, location-privacy preserving dashcam
online crowdsourcing mechanism. In: Proceedings of the 2018 2nd International video sharing using blockchain and deep learning. Hum.-centric Comput. Inf. Sci.
Conference on Computer Science and Artificial Intelligence. pp. 195–199. 10 (1), 1–23.
Guegan, D., 2017. Public blockchain versus private blockhain. Kirillova, E., Bogdan, V.V., Filippov, P., Tkachev, V., Zulfugarzade, T., 2020. The
Guo, H., Dolhansky, B., Hsin, E., Dinh, P., Ferrer, C.C., Wang, S., 2021. Deep poisoning: main features of blockchain technologies classification. COMPUSOFT: Int. J. Adv.
Towards robust image data sharing against visual disclosure. In: Proceedings of the Comput. Technol. 9 (10), 3900–3905.
IEEE/CVF Winter Conference on Applications of Computer Vision. pp. 686–696. Kumar, R., Wang, W., Kumar, J., Yang, T., Khan, A., Ali, W., Ali, I., 2021. An integration
Gupta, S., Hellings, J., Rahnama, S., Sadoghi, M., 2020. Building high throughput of blockchain and AI for secure data sharing and detection of CT images for the
permissioned blockchain fabrics: challenges and opportunities. Proc. VLDB Endow. hospitals. Comput. Med. Imaging Graph. 87, 101812.
13 (12), 3441–3444. Lee, X.T., Khan, A., Sen Gupta, S., Ong, Y.H., Liu, X., 2020. Measurements, analyses,
Gupta, R., Kumari, A., Tanwar, S., 2021. Fusion of blockchain and artificial intelli- and insights on the entire ethereum blockchain network. In: Proceedings of the
gence for secure drone networking underlying 5G communications. Trans. Emerg. Web Conference 2020. pp. 155–166.
Telecommun. Technol. 32 (1), e4176. Lee, D., Park, N., 2020. Blockchain based privacy preserving multimedia intelligent
Hadid, A., Evans, N., Marcel, S., Fierrez, J., 2015. Biometrics systems under spoofing video surveillance using secure Merkle tree. Multimedia Tools Appl. 1–18.
attack: an evaluation methodology and lessons learned. IEEE Signal Process. Mag. Li, R., 2020. Fingerprint-related chaotic image encryption scheme based on blockchain
32 (5), 20–30. framework. Multimedia Tools Appl. 1–21.
Hafiz, A., Hassaballah, M., Alqahtani, A., Alsubai, S., Hameed, M.A., 2023. Reinforce- Li, Y., Tu, Y., Lu, J., Wang, Y., 2020. A security transmission and storage solution
ment learning with an ensemble of binary action deep Q-networks. Comput. Syst. about sensing image for blockchain in the Internet of Things. Sensors 20 (3), 916.
Sci. Eng. 46 (3). Liu, M., Shang, J., Liu, P., Shi, Y., Wang, M., 2018. VideoChain: trusted video
Haiwu, H., An, Y., Zehua, C., 2018. Survey of smart contract technology and application surveillance based on blockchain for campus. In: International Conference on Cloud
based on blockchain. J. Computer Research and Development 55 (11), 2452. Computing and Security. Springer, pp. 48–58.
Hameed, M.A., Abdel-Aleem, O.A., Hassaballah, M., 2023. A secure data hiding Liu, M., Teng, Y., Yu, F.R., Leung, V.C., Song, M., 2020a. A deep reinforcement
approach based on least-significant-bit and nature-inspired optimization techniques. learning-based transcoder selection framework for blockchain-enabled wireless D2D
J. Ambient Intell. Humaniz. Comput. 14 (5), 4639–4657. transcoding. IEEE Trans. Commun. 68 (6), 3426–3439.
Hasan, H.R., Salah, K., 2019. Combating deepfake videos using blockchain and smart Liu, M., Teng, Y., Yu, F.R., Leung, V.C., Song, M., 2020b. A mobile edge computing
contracts. IEEE Access 7, 41596–41606. (MEC)-enabled transcoding framework for blockchain-based video streaming. IEEE
Hassaballah, M., Aly, S., Abdel Rady, A.S., et al., 2018. A high payload steganography Wirel. Commun. 27 (2), 81–87.
method based on pixel value differencing. Lopes, V., Pereira, N., Alexandre, L.A., 2019. Robot workspace monitoring using a
Hassaballah, M., Hameed, M.A., Alkinani, M.H., 2020a. Introduction to digital image blockchain-based 3D vision approach. In: Proceedings of the IEEE/CVF Conference
steganography. In: Digital Media Steganography. Elsevier, pp. 1–15. on Computer Vision and Pattern Recognition Workshops.
Hassaballah, M., Hameed, M.A., Aly, S., AbdelRady, A., 2020b. A color image Lumini, A., Nanni, L., 2006. An advanced multi-modal method for human authentica-
steganography method based on ADPVD and HOG techniques. In: Digital Media tion featuring biometrics data and tokenised random numbers. Neurocomputing 69
Steganography. Elsevier, pp. 17–40. (13–15), 1706–1710.
23
Mayer, O., Stamm, M.C., 2020. Exposing fake images with forensic similarity graphs. Sathish, S.K., Patankar, A.A., Khanna, H., 2019. Aurum: A blockchain based decen-
IEEE J. Sel. Top. Sign. Proces. 14 (5), 1049–1064. tralized video streaming platform. In: 2019 IEEE Wireless Communications and
McConaghy, T., Marques, R., Müller, A., De Jonghe, D., McConaghy, T., Mc- Networking Conference. WCNC, IEEE, pp. 1–8.
Mullen, G., Henderson, R., Bellemare, S., Granzotto, A., 2016. Bigchaindb: A Sawant, G., Bharadi, V., 2020. Permission blockchain based smart contract utilizing
Scalable Blockchain Database. White Paper, BigChainDB. biometric authentication as a service: A future trend. In: 2020 International
Mehta, R., Kapoor, N., Sourav, S., Shorey, R., 2019. Decentralised image sharing Conference on Convergence to Digital World-Quo Vadis. ICCDW, IEEE, pp. 1–4.
and copyright protection using blockchain and perceptual hashes. In: 2019 11th Seo, J., Cho, Y., 2020. Medical image sharing system using hyperledger fabric
International Conference on Communication Systems & Networks. COMSNETS, blockchain. In: 2020 22nd International Conference on Advanced Communication
IEEE, pp. 1–6. Technology. ICACT, IEEE, pp. 62–64.
Mendki, P., 2019. Blockchain enabled IoT edge computing. In: Proceedings of the 2019 Shafique, K., Khawaja, B.A., Sabir, F., Qazi, S., Mustaqim, M., 2020. Internet of things
International Conference on Blockchain Technology. pp. 66–69. (IoT) for next-generation smart systems: A review of current challenges, future
trends and prospects for emerging 5G-IoT scenarios. IEEE Access 8, 23022–23040.
Michelin, R.A., Ahmed, N., Kanhere, S.S., Seneviratne, A., Jha, S., 2020. Leveraging
Shalaby, S., Abdellatif, A.A., Al-Ali, A., Mohamed, A., Erbad, A., Guizani, M., 2020. Per-
lightweight blockchain to establish data integrity for surveillance cameras. In: 2020
formance evaluation of hyperledger fabric. In: 2020 IEEE International Conference
IEEE International Conference on Blockchain and Cryptocurrency. ICBC, IEEE, pp.
on Informatics, IoT, and Enabling Technologies. ICIoT, IEEE, pp. 608–613.
1–3.
Shen, M., Deng, Y., Zhu, L., Du, X., Guizani, N., 2019. Privacy-preserving image
Mittal, V., Tyagi, A., Bhushan, B., 2020. Smart surveillance systems with edge
retrieval for medical IoT systems: A blockchain-based approach. IEEE Network 33
intelligence: Convergence of deep learning and edge computing. Available at SSRN
(5), 27–33.
3599865.
Shi, W., Cao, J., Zhang, Q., Li, Y., Xu, L., 2016. Edge computing: Vision and challenges.
Mohanta, B.K., Panda, S.S., Jena, D., 2018. An overview of smart contract and use cases IEEE Internet Things J. 3 (5), 637–646. http://dx.doi.org/10.1109/JIOT.2016.
in blockchain technology. In: 2018 9th International Conference on Computing, 2579198.
Communication and Networking Technologies. ICCCNT, IEEE, pp. 1–4. Shrier, D., Wu, W., Pentland, A., 2016. Blockchain & infrastructure (identity, data
Mohsin, A., Zaidan, A., Zaidan, B., Albahri, O., Albahri, A., Alsalem, M., Mohammed, K., security). Mass. Inst. Technol.-Connect. Sci. 1 (3), 1–19.
2019. Based blockchain-PSO-AES techniques in finger vein biometrics: A novel Singh, A.K., 2020. A multi-layered network model for blockchain based security
verification secure framework for patient authentication. Comput. Stand. Inter- surveillance system. In: 2020 IEEE International Conference for Innovation in
faces 66, 103343. http://dx.doi.org/10.1016/j.csi.2019.04.002, URL: https://www. Technology. INOCON, IEEE, pp. 1–5.
sciencedirect.com/science/article/pii/S0920548918304793. Singh, M., Aujla, G.S., Bali, R.S., 2020. Odob: One drone one block-based lightweight
Mudliar, K., Parekh, H., Bhavathankar, P., 2018. A comprehensive integration of blockchain architecture for internet of drones. In: IEEE INFOCOM 2020-IEEE
national identity with blockchain technology. In: 2018 International Conference on Conference on Computer Communications Workshops. INFOCOM WKSHPS, IEEE,
Communication Information and Computing Technology. ICCICT, IEEE, pp. 1–6. pp. 249–254.
Naganuma, K., Suzuki, T., Yoshino, M., Takahashi, K., Kaga, Y., Kunihiro, N., 2020. New Smith, D.F., Wiliem, A., Lovell, B.C., 2015. Face recognition on consumer devices:
secret key management technology for blockchains from biometrics fuzzy signature. Reflections on replay attacks. IEEE Trans. Inf. Forensics Secur. 10 (4), 736–745.
In: 2020 15th Asia Joint Conference on Information Security. AsiaJCIS, IEEE, pp. http://dx.doi.org/10.1109/TIFS.2015.2398819.
54–58. Sreenu, G., Durai, M.S., 2019. Intelligent video surveillance: a review through deep
Nagothu, D., Xu, R., Nikouei, S.Y., Chen, Y., 2018. A microservice-enabled architecture learning techniques for crowd analysis. J. Big Data 6 (1), 1–27.
for smart surveillance using blockchain technology. In: 2018 IEEE International Sunyaev, A., 2020. Distributed ledger technology. In: Internet Computing. Springer, pp.
Smart Cities Conference. ISC2, IEEE, pp. 1–4. 265–299.
Tang, H., Tong, N., Ouyang, J., 2018. Medical images sharing system based on
Nakamoto, S., Bitcoin, A., 2008. A peer-to-peer electronic cash system, 4. Bitcoin.–URL:
blockchain and smart contract of credit scores. In: 2018 1st IEEE International
https://bitcoin.org/bitcoin.pdf.
Conference on Hot Information-Centric Networking. HotICN, IEEE, pp. 240–241.
Nikouei, S.Y., Xu, R., Nagothu, D., Chen, Y., Aved, A., Blasch, E., 2018. Real-time
Taş, R., Tanrıöver, Ö.Ö., 2019. Building a decentralized application on the ethereum
index authentication for event-oriented surveillance video query using blockchain.
blockchain. In: 2019 3rd International Symposium on Multidisciplinary Studies and
In: 2018 IEEE International Smart Cities Conference. ISC2, IEEE, pp. 1–8.
Innovative Technologies. ISMSIT, IEEE, pp. 1–4.
Nyaletey, E., Parizi, R.M., Zhang, Q., Choo, K.-K.R., 2019. BlockIPFS-blockchain-enabled
Toutara, F., Spathoulas, G., 2020. A distributed biometric authentication scheme based
interplanetary file system for forensic and trusted data traceability. In: 2019 IEEE
on blockchain. In: 2020 IEEE International Conference on Blockchain. Blockchain,
International Conference on Blockchain. Blockchain, IEEE, pp. 18–25.
IEEE, pp. 470–475.
Othman, A., Callahan, J., 2018. The horcrux protocol: a method for decentralized Tsakanikas, V., Dagiuklas, T., 2018. Video surveillance systems-current status and future
biometric-based self-sovereign identity. In: 2018 International Joint Conference on trends. Comput. Electr. Eng. 70, 736–753.
Neural Networks. IJCNN, IEEE, pp. 1–7. Vukolić, M., Hyperledger fabric. Genesis 1.
Paralkar, K., Yadav, S., Kumari, S., Kulkarni, A., Pingat, S., 2018. Photogroup: Wang, R., Tsai, W.-T., He, J., Liu, C., Li, Q., Deng, E., 2019. A video surveillance
Decentralized web application using ethereum blockchain. Int. Res. J. Eng. Technol. system based on permissioned blockchains and edge computing. In: 2019 IEEE
5, 489–492. International Conference on Big Data and Smart Computing. BigComp, IEEE, pp.
Patel, D., Mistry, V., et al., 2018. Border control and immigration on blockchain. In: 1–6.
International Conference on Blockchain. Springer, pp. 166–179. Wu, H., Ashikhmin, A., Wang, X., Li, C., Yang, S., Zhang, L., 2020. Distributed error
Patel, P., Thakkar, A., 2020. The upsurge of deep learning for computer vision correction coding scheme for low storage blockchain systems. IEEE Internet Things
applications. Int. J. Electr. Comput. Eng. 10 (1), 538. J. 7 (8), 7054–7071.
Peck, M.E., 2017. Blockchain world-do you need a blockchain? This chart will tell you Xu, Q., He, Z., Li, Z., Xiao, M., 2018. Building an ethereum-based decentralized
if the technology can solve your problem. IEEE Spectr. 54 (10), 38–60. smart home system. In: 2018 IEEE 24th International Conference on Parallel and
Peixoto, J.P.J., Costa, D.G., 2017. Wireless visual sensor networks for smart city Distributed Systems. ICPADS, IEEE, pp. 1004–1009.
applications: A relevance-based approach for multiple sinks mobility. Future Gener. Xu, R., Nikouei, S.Y., Chen, Y., Blasch, E., Aved, A., 2019. Blendmas: A blockchain-
Comput. Syst. 76, 51–62. enabled decentralized microservices architecture for smart public safety. In: 2019
IEEE International Conference on Blockchain. Blockchain, IEEE, pp. 564–571.
Pongnumkul, S., Siripanpornchana, C., Thajchayapong, S., 2017. Performance analysis
Yasaweerasinghelage, R., Staples, M., Weber, I., 2017. Predicting latency of blockchain-
of private blockchain platforms in varying workloads. In: 2017 26th International
based systems using architectural modelling and simulation. In: 2017 IEEE
Conference on Computer Communication and Networks. ICCCN, IEEE, pp. 1–6.
International Conference on Software Architecture. ICSA, IEEE, pp. 253–256.
Predd, J.B., Kulkarni, S.B., Poor, H.V., 2006. Distributed learning in wireless sensor
Yatskiv, V., Yatskiv, N., Bandrivskyi, O., 2019. Proof of video integrity based
networks. IEEE Signal Process. Mag. 23 (4), 56–69.
on blockchain. In: 2019 9th International Conference on Advanced Computer
Qi, X., Yu, K., Wen, Z., Katsuyama, Y., Sato, T., Tokuda, K., Sato, T., et al., Information Technologies. ACIT, IEEE, pp. 431–434.
2020. Blockchain-based content-oriented surveillance network. In: 2020 IEEE 91st Youssef, S.B.H., Rekhis, S., Boudriga, N., 2019. A blockchain based secure IoT solution
Vehicular Technology Conference. VTC2020-Spring, IEEE, pp. 1–6. for the dam surveillance. In: 2019 IEEE Wireless Communications and Networking
Ramachandran, G.S., Krishnamachari, B., 2018. Blockchain for the IoT: Opportunities Conference. WCNC, IEEE, pp. 1–6.
and challenges. arXiv preprint arXiv:1805.02818. Zelensky, A., Voronin, V., Semenishchev, E., Svirin, I., Alepko, A., 2018. Video content
Rebecq, H., Ranftl, R., Koltun, V., Scaramuzza, D., 2019. Events-to-video: Bringing verification using blockchain technology. In: 2018 IEEE International Conference
modern computer vision to event cameras. In: Proceedings of the IEEE/CVF on Smart Cloud. SmartCloud, IEEE, pp. 208–212.
Conference on Computer Vision and Pattern Recognition. pp. 3857–3866. Zerka, F., Urovi, V., Vaidyanathan, A., Barakat, S., Leijenaar, R.T., Walsh, S., Gabrani-
Reddy, R.V., prakash Reddy, V.J., Reddy, E.M., 2020. Automatic face expressions Juma, H., Miraglio, B., Woodruff, H.C., Dumontier, M., et al., 2020. Blockchain for
and gesture detection system using blockchain security. In: 2020 International privacy preserving and trustworthy distributed machine learning in multicentric
Conference on Intelligent Engineering and Management. ICIEM, IEEE, pp. 296–300. medical imaging (C-DistriM). IEEE Access 8, 183939–183951.
24
Zhang, D., Guo, Z., Lu, G., Zhang, L., Liu, Y., Zuo, W., 2011. Online joint palmprint Zikratov, I., Kuzmin, A., Akimenko, V., Niculichev, V., Yalansky, L., 2017. Ensuring
and palmvein verification. Expert Syst. Appl. 38 (3), 2621–2631. data integrity using blockchain technology. In: 2017 20th Conference of Open
Zheng, X., Mukkamala, R.R., Vatrapu, R., Ordieres-Mere, J., 2018. Blockchain-based Innovations Association. FRUCT, IEEE, pp. 534–539.
personal health data sharing system using cloud storage. In: 2018 IEEE 20th Inter- Zupan, N., Zhang, K., Jacobsen, H.A., 2017. Hyperpubsub: a decentralized, permis-
national Conference on E-Health Networking, Applications and Services. Healthcom, sioned, publish/subscribe service using blockchains. In: Proceedings of the 18th
IEEE, pp. 1–6. ACM/IFIP/USENIX Middleware Conference: Posters and Demos. pp. 15–16.
25

Ejemplo - Block Chain

Uploaded by

Copyright:

Available Formats

Ejemplo - Block Chain

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ejemplo - Block Chain

Uploaded by

Copyright:

Available Formats

Journal of King Saud University - Computer and Information Sciences 35 (2023) 101801

Contents lists available at ScienceDirect

Journal of King Saud University - Computer and Information

Enhanced computer vision applications with blockchain: A review of

ARTICLE INFO ABSTRACT

Fig. 1. Sensor network in surveillance systems.

Fig. 2. Overview of common computer vision applications with blockchain.

entities were given tokens based on privilege of access. The application

Fig. 4. Overview of surveillance system with blockchain.

secret scheme which are then stored using a multi-distribution scheme.

Fig. 8. Typical video streaming process.

Fig. 9. Medical sharing between organization .

Recent developments in hardware and embedded computing has

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.