21CSE252T- BIOMETRICS

Elective
Academic Year 2025 (Even Semester)

Dr. B. AMUTHA
Professor/CTech

Mobile: 9445242714
Email: amuthab@srmist.edu.in
Room: TP106

Cloud Computing 1
 GCR Details

GCR Link

https://classroom.google.com/c/NzMzODYyMzk3MTQx?cjc=uidkktf

GCR Code
uidkktf

2
 Course Outcomes (COs):

The purpose of learning this course is to

Cloud Computing 3
 Learning Outcomess will
At the end of this course, learners will be able to:
• Acquire the knowledge on various biometric traits
• Acquire the ability to identify pattern recognition
system and its features
• Understand the basic ideas about physical and
behavioral biometric traits
• Apply the knowledge of biometrics on developing
identification system.
• Apply the knowledge for designing biometric
systems
• Acquire the knowledge on authentication systems
for real time security applications
Cloud Computing 4
CO-PO Mappings

Cloud Computing 5
Overall Assessment Plan

Cloud Computing 6
Learning Resources

Cloud Computing 7
 Learning Resources

Reference Books

 Introduction to Biometrics – James Wayman, -

Springer
 Feature Extraction and Image Processing
for Computer Vision – Mark S Necon – Elsevir
 Digital Image Processing using Mtalab –
Rafael C, - Tata MC Graw Hill
 Guide to Biometrics – Rood M Bolie –Springer
 Pattern Classification – Richard Oduda –
Wiley
 Biometrics in Identity Management:
Concepts to Applications – Shimon K Modi
Cloud Computing 8
Unit – 1

CLR1 : Understanding the concept of Authentication using Biometrics

Cloud Computing 9
UNIT – I Contents
• Introduction to Biometric Systems
• Biometric Functionalities :Verification and
Identification
• The Design Cycle of Biometric Systems
• Introduction to Unimodel Systems
• Introduction to multi-model Biometric Systems
• Biometric System Errors & Performance
Measures
UNIT – I Contents
• Image Processing Basics
• First & Second Derivatives
• Steps in Edge Detection, Smoothing, enhancement,
etc
• Robert’s method, Sobel’s method and Perwits
Method
• Laplacian of Gaussian, Zero Crossing
• Low Level Feature Extraction, Describing Image
Motion
• High Level Feature Extraction, Template Matching
• Hough Transforms for Lines, Circles and Ellipses
 Session – 1 & 2
• Introduction to Biometric Systems
– Biometric Functionalities :Verification and
Identification

• The design cycle of biometric systems

– The building blocks of generic biometric systems
Introduction to Biometric Systems
Introduction to Biometric Systems

• What does biometrics mean?

• Why biometrics?
• What is biometrics used for
• What are the types of biometrics?
• Who invented biometrics? (History of
biometrics)
• Is biometrics accurate and reliable
Introduction to Biometric Systems

• Biometrics can be defined as the means of

identifying and authenticating individuals in a
reliable and fast way through unique biological
characteristics.
• It is the science of establishing the identity of an
individual based on the physical, chemical or
behavioural attributes of the person.
• Biometrics allows a person to be identified and
authenticated based on recognizable and
verifiable data, unique and specific.
 Why Biometrics?
• Needs: For large-scale identity management systems
whose functionality relies on the accurate determination
of an individual's identity in the context of several
different applications
• The proliferation of web-based services (e.g., online
banking)
• The deployment of decentralized customer service
centres (e.g., credit cards)
• The main aim is to prevent impostors from accessing
protected resources
 History of Biometrics
• In the second century B.C., the Chinese emperor Ts'In
was a authenticating specific seals with a fingerprint.
• Fingerprints were first used in a commercial setting in
1858 by William James Herschel, a British administrator
in India. Having been put in charge of building roads in
Bengal, he had his subcontractors sign contracts with
their fingers.
• At the end of the 19th century, Bertillon, a French police
officer, took the first steps in scientific policing.
• He used physical measurements of specific anatomical
characteristics to identify reoffending criminals
 Types of Biometrics
There are two types of biometrics:
#1. Physiological measurements
• They can be either morphological or biological.
• Morphological identifiers mainly consist of
fingerprints, the hand's shape, the finger, vein
pattern, the eye (iris and retina), and the face's
shape.
• For biological analyses, DNA, blood, saliva, or
urine may be used by medical teams and police
forensics.
 Types of Biometrics
#2. Behavioral measurements
The most common are:
• voice recognition,
• signature dynamics (speed of movement of pen,
accelerations, pressure exerted, inclination),
• keystroke dynamics,
• the way we use objects,
• gait, the sound of steps,
• gestures, etc.
Types of Biometrics
Traditional Methods and Biometrics
• Traditional methods of establishing a person's identity
– Knowledge- based (e.g., passwords) and
– Token-based (e.g., ID cards) mechanisms
• Drawbacks: Representations of identity can easily
be lost, shared, manipulated or stolen thereby
compromising the intended security.
• Dual-factor Authentication Scheme:
• Biometrics is used to supplement ID cards and
passwords thereby imparting an additional level
of security.
 Biometrics for Identity
Person Recognition in Identity Management
• The aim is to establish the association between an
individual and his personal identity.
• One must be able to determine a person’s identity or
verify the identity claim of an individual whenever
required
• This process is known as person recognition.
• A person can be recognized based on
(a) what he knows,
(b) what he possesses extrinsically, and
(c) who he is intrinsically (biometric recognition)
 Biometrics for Identity

• The three different types of authentication

– Something you know [knowledge based
system]
• password, PIN
– Something you have [token based system]
• Card key, smart card

– Something you are Biometric

 Attacks in Authentication Systems
Various types of Malicious Attacks on Authentication
Systems
(a) client attack (e.g. guessing passwords, stealing tokens);
(b) host attack (e.g. accessing plain text ¯le containing
passwords);
(c) eavesdropping (e.g. shoulder surfing for passwords);
(d) repudiation (e.g., claiming that token was misplaced);
(e) trojan horse attack (e.g., installation of bogus log-in screen
to steal passwords); and
(f) denial of service (e.g., disabling the system by deliberately
supplying an incorrect password several times).
Biometric Systems Advantages and
Disadvantages
Advantages
• Uniqueness
• No need to remember password and pin
• Cannot be lost, stolen or forgotten
• Difficult to copy, share and distribute
• Person is required to be present for authentication

Disadvantages
• Violation of Privacy
• Need of significant computational resources
• Intra-class variation: Due to change in pose or age
• Vulnerable to spoof attacks
 Traits, indicators, identifiers or
modalities.
Physical/Behavioural Characteristics used in BS
• Fingerprint, face, hand/Finger geometry, iris,
retina, signature, gait, palm-print, voice pattern,
ear, hand vein, odour or the DNA information of
an individual to establish identity
• These characteristics are referred to as traits,
indicators, identifiers or modalities.
Traits, indicators, identifiers or
modalities.
Different Traits
Operation of a Generic Biometric
System
A generic biometric system

Database
Enrollment

Matching
Feature Enroll/
Preprocessing
Extraction Match

Data Acquisition Verification/

(Sensor/Camera) Identification
 Operation of a Biometric System
• A Biometric System is a pattern recognition system
that acquires biometric data from an individual,
extracts a salient feature set from the data, compares
this feature set against the feature set(s) stored in the
database, and executes an action based on the result
of the comparison.
The main modules are
• a sensor module;
• a quality assessment
• feature extraction module;
• a matching module; and
• a database module.
Detailed Building Blocks of Generic
Biometric Systems
 Operation of a Biometric System
• Depending upon the application context, system operates in
– Verification Mode: (1:1)
– Identification Mode: (1:N)
 Operation of a Biometric System
Verification Mode
• The system validates a person’s identity by comparing the
captured biometric data with her own biometric template
stored system database (one-to-one matching)
• Verification asks “Is the person who they say they are?”
• Eg: Does the biometric data belong to Ajay?
Identification mode
• The system recognizes an individual by searching the
template of all the users in the database for a match. (one-
to-many)
• Identification is the task of answering: “Who is this person?”
• Eg: whose biometric data is this?
 Sensor Module
• A suitable biometric reader or scanner is required to
acquire the raw biometric data of an individual.
• To obtain Fingerprint images, for example, an optical
Fingerprint sensor may be used to get the friction ridge
structure of the Fingertip.
• The sensor module defines the human machine
interface
 Quality assessment and feature
extraction module
• The sensor data is assessed in order to determine its
suitability for further processing.
• Typically, the acquired data is subjected to a signal
enhancement algorithm in order to improve its quality.
• A set of salient discriminatory features are extracted to
represent the underlying trait. For example, the position
and orientation of minutia points (local ridge and valley
anomalies) in a fingerprint image are extracted by the
feature extraction module in a fingerprint-based bio
metric system.
• During enrolment, this feature set is stored in the
database and is commonly referred to as a template
Quality assessment and feature
extraction module
 Database Module
• The database acts as the repository of bio-metric
information.
• Raw biometric sample (i.e., the template) is stored in the
database along with some biographic information
characterizing the user
• Some systems store multiple templates in order to
account for the intra-class variations associated with a
user.
• Face recognition systems, for instance, may store
multiple templates of an individual, with each template
corresponding to a different facial pose with respect to
the camera.
 Database Module

• The raw biometric images may also be stored in

the database along with the templates during
enrolment. Such images are often known as
gallery images, reference images, stored
images, or enrolment images.
• The images acquired during recognition are
known as probe images, query images, or input
images
 Matching and Decision-Making
Module:
• The extracted features are compared against the stored
templates to generate match scores.
• In a fingerprint-based biometric system, the number of
matching minutiae between the input and the template
feature sets is determined and a match score reported.
• The match score may be moderated by the quality of the
presented biometric data.
• The matcher module also encapsulates a decision
making module, in which the match scores are used to
either validate a claimed identity or provide a ranking of
the enrolled identities in order to identify an individual.
Biometric Functionalities
 Biometric Functionalities
• A biometric system can provide two types of identity management
functionalities, namely, verification and identification
Verification
• Here, the user claims an identity and the system verifies whether
the claim is genuine, i.e., the system answers the question “Are
you who you say you are?”.
• In this scenario, the query is compared only to the template
corresponding to the claimed identity (a one-to-one match).
• Verification is typically used in applications where the goal is to
prevent unauthorized persons from using the services
• Verification as the two-category classification problem:
• Given a claimed identity I and a query feature set xA, we need to
decide if (I,xA) belongs to “genuine” or “impostor” class.
 Biometric Functionalities
• Let x Ͼ I be the stored template corresponding to identity
I. Typically, xA is compared with x Ͼ I and a match score s,
which measures the similarity between xA and x Ͼ I , is
computed.
• The decision rule is given by

• where η is a pre-defined threshold

• When the identity claim is deemed to be “genuine”, the
user is allowed to access the services provided by the
system
 Biometric Functionalities
Identification
• Identification functionality can be further classified into positive
and negative identification.
• In positive identification, the user attempts to positively identify
himself to the system without explicitly claiming an identity.
• A positive identification system answers the question “Are you
someone who is known to the system?” by determining the
identity of the user from a known set of identities.
• In contrast, the user in a negative identification application is
considered to be concealing his true identity (either explicitly or
implicitly) from the system.
• Negative identification is also known as screening and the
objective of such systems is to find out “Are you who you say you
are not?”.
 Biometric Functionalities
• The purpose of negative identification is to prevent a
single person from using multiple identities.
• Hence, screening can be used to prevent the issue of
multiple credential records (e.g., driver’s licence,
passport) assigned to the same person or to prevent a
person from claiming multiple benefits under different
names
• In both positive and negative identification, the user’s
biometric input is compared with the templates of all the
persons enrolled in the database and the system outputs
either the identity of the person whose template has the
highest degree of similarity with the user’s input.
 Biometric Functionalities
Differences between “Positive” & “Negative” Identification
 Biometric Functionalities
• Formally, the problem of identification can be stated as follows:
• Given a query feature set xA, we need to decide the identity I of
the user, where I ∈ {I1, I2, · · · , IN, IN+1}. Here, I1, I2, · · · , IN
correspond to the identities of the N users enrolled in the system
and IN+1 indicates the case where no suitable identity can be
determined for the given query. If x Ͼ In is the stored template
corresponding to identity In and sn is the match score between xA
and xEIn, for n = 1,2, · · · ,N, the decision rule for identification is,

• where η is a pre-defined threshold.

• The above decision rule is commonly known as open set
identification
 Biometric Functionalities
• It is also possible to force the system to return one
among the N enrolled identities, irrespective of the value
of sn0 . Such a scenario is called closed set identification.
• In some practical biometric identification systems (e.g.,
latent fingerprint matching), identification is semi-
automated.
• A semi-automated biometric system outputs the
identities of the top t matches (1 < t Ͼ N) and a human
expert manually determines the identity (among the t
selected identities) that best matches the given query.
• The value of t could be determined based on the
availability and throughput of the human expert(s).
 The Design Cycle of Biometric
Systems
Major activities in the design of BS (Iterative)
• Understanding the nature of the application and
the performance requirements
• Choosing the right biometric trait(s) for the
application in hand
• Collect biometric data from a subset of target
population
• Design or train the core biometric modules,
including the feature extractor and the matcher
• Evaluation procedure to ensure that it meets the
requirements of the application
The Design Cycle of Biometric
Systems
 A Taxonomy of Application
Environments
• The variations in the application environment had
a significant impact on the way the devices
performed.
• In fact, accurate characterization of the
operational environment is primary in selecting
the best biometric technology and in predicting
the system’s operational characteristics.
• Depending on the application context, we may
need to choose between the verification and
identification functionalities
Factors influencing the Nature of the
Application
1. Nature of the application
• Biometric applications can also be classified based on the
following issues (Nature of the App).
– Cooperative versus non-cooperative user
– Overt versus covert deployment
– Habituated users versus non-habituated users
– Attended versus unattended operation
– Controlled versus uncontrolled operation
– Open versus closed system
Factors influencing the Nature of the
Application
• Cooperative versus non-cooperative users
• Eg. (Cooperative) - E-Banking
• Eg. (Non-cooperative) – Terrorist’s behaviours in airport
screening application
• Overt versus Covert deployment
– If the user is aware that he is being subjected to
biometric recognition, the application is categorized as
overt (Finger Print Verification System)
– If the user is unaware, the application is called covert
(Facial Recognition)
Factors influencing the Nature of the
Application
• Habituated users versus non-habituated users
– If the enrolled users interact with the biometric system
quite frequently, they tend to get habituated in
providing their biometric data. (Login App in Computer
Network)
– A driver’s license application typically has non-
habituated users since a driver’s license is renewed
only once in a period of several years.
– The familiarity of users with the system can affect
recognition accuracy
Factors influencing the Nature of the
Application
• Attended versus unattended operation
– It refers to whether the process of biometric data
acquisition in an application is observed, guided, or
supervised by a human (e.g., a security officer).
– An application may have an attended enrollment
operation but unattended recognition operation
– For example, a banking application may have a
supervised enrollment when an ATM card is issued to a
user, but the subsequent uses of the biometric system
for the ATM transaction are not attended
Factors influencing the Nature of the
Application
• Controlled versus uncontrolled operation: In a
controlled environment, ambient environmental
conditions such as temperature, pressure, moisture,
lighting conditions, etc. can be moderated during the
operation of a biometric system.
• Typically, indoor applications such as computer network
login operate in a controlled environment, whereas
outdoor applications such as keyless car entry or parking
lot surveillance operate in an uncontrolled environment.
• This classification is also important for the system
designer as a more rugged biometric sensor is needed
for an uncontrolled environment
Factors influencing the Nature of the
Application
• Open versus closed system: If a person’s biometric
template can be used across multiple applications, the
biometric system can be considered as open.
• For example, a user may use a fingerprint-based
recognition system for entering secure facilities, computer
network login, electronic banking, and bank ATMs.
• When all these applications use separate templates
(databases) for each application, the system is considered
closed.
• A closed system may be based on a proprietary template
whereas an open system will need standard data formats
and data compression methods to exchange and compare
information between different systems
 Factors influencing the Biometric
Traits
2. Choice of biometric trait
• Each biometric trait has its pros and cons and,
therefore, the choice of a biometric trait for a
particular application depends on a variety of
issues.
• In general, seven factors must be considered to
determine the suitability of a physical or a
behavioural trait to be used in a biometric
application.
Factors influencing the Biometric Traits
1. Universality: We should be able to find our chosen
biometric characteristic in the majority of people we expect
to enrol in the system. For instance, although we might be
able to use a scar as an identifier, we cannot guarantee that
everyone will have a scar. It determines the failure to enroll
(FTE) rate of the biometric system
2. Uniqueness: The given trait should be different across
individuals comprising the user population. Otherwise, the
false match rate (FAR or FPIR) of the biometric system
would be high
3. Permanence: The biometric trait of an individual should be
invariant over a period of time with respect to matching
algorithm, else, it will lead to a high false non-match rate
Factors influencing the Biometric Traits

4. Measurability: It should be possible to acquire

and digitize the biometric trait using suitable devices
that do not cause inconvenience to the individual.
This factor significantly impacts the frequency of FTE
and FTA failures and the recognition accuracy
5. Performance: The computational resources
required to achieve that accuracy and throughput
requirements
6. Acceptability: Individuals in the target population
that will utilize the application should be willing to
present their biometric trait to the system
Factors influencing the Biometric Traits

7. Circumvention: This refers to the ease with which

the trait of an individual can be imitated using
artifacts (e.g., fake fingers), in the case of physical
traits, and mimicry, in the case of behavioural traits.
It also refers to the process of obfuscation, where a
user deliberately alters his biometric trait to evade
recognition
No single biometric is expected to effectively meet
all the requirements (e.g., accuracy, practicality,
cost) imposed by all applications (e.g., forensics,
access control, etc.)
 Commonly used biometric
characteristics
• Fingerprint
• Palm print
• Iris
• Face
• Hand Geometry (Shape)
• Gait
• Ear
• Voice
• Keystroke
• Signature
• DNA
• Facial, hand, and hand vein infrared thermograms
• Odour
Types of Commonly used Biometric
Characteristics
Taxonomy of Commonly used
Biometric Characteristics
The advantages of Physiological based biometric
authentication mechanisms
Physiological biometrics:
• It does not take more than a few seconds to authenticate
a person’s identity by fingerprint, face or voice.
• One does not need to remember complex passwords,
keys, tokens or smart cards to validate his identity.
• The characteristics of this mechanism do not change e.g.
Iris, fingerprint, DNA etc.
• 99% accuracy and reliability are being provided via
distinctness in recognition.
• Social acceptance of using this mechanism is high.
• Hacking chances via using this are very less.
The advantages of using Behavioral based
biometric authentication mechanisms
Behavioral biometrics:
• Behavioral biometric authenticate the users without the
need to do some specific action.
• It collects the data for authentication dynamically.
• By using it, user can maintain privacy and avoid social
awkwardness.
• It prevent identity theft and minimize the risk of online
fraud.
• It is very authentic scheme and based on the user
experience and individual skills.
• 90% accuracy is provided by this in recognition
 The disadvantages of using these biometric
authentication mechanisms
Physiological biometrics:
• Physiological Biometric authentication equipment is
costly.
• People have to wait in line to get scanned to gain
access which can cause delays.
• Slightly change in facial expression or obstruction
due to hat, glasses or your finger is hurt or the voice
is affected by cold couldn’t be recognized by the
system.
• Dry, wet or dirty fingers can oscillate performance
 The disadvantages of using these biometric
authentication mechanisms
Behavioral biometrics:
• Behavioral biometric data can be used by people who
want to steal them.
• It has some effects on the privacy of individuals.
• Sometimes have problem in input methods such as
phonetic may suffer from language problem.
• It would be difficult to identify the gesture if the
movement slightly varies.
• It provides the less reliability than physiological bio
metric behavior
 The rationale for choosing Behavioral vs
physiological biometric authentication mechanisms
Behavioral biometrics:
• Behavioral biometric authentication is preferred
because the behavioral biometric system database is
dynamic, making it more secure.
• It has less chance of being copied, stolen or reused by
unauthorized personnel, unlike physiological
biometrics.
• It does not require any specialized hardware that
makes it economic friendly. These characteristics
make behavioral biometrics a good option in high-
security situation
 The rationale for choosing Behavioral vs
physiological biometric authentication mechanisms

Physiological biometrics:
• Physiological biometrics are hardware-dependent, unlike
biometric behavior.
• Physiological biometric authentication provides accurate
results compared to the behavioral mechanism.
Keeping these key points in mind, it is a better alternative
to use physiological biometrics if the behavioral mechanism
does not authenticate the user.
Physiological biometrics should be sufficient in
environments where high protection is not required
The social issues in Biometric Authentication
mechanism
• The biometric-based authentication system can be
attacked by Insider attacker and stranger
• A scammer can make various attempts to spoof a
legitimate user's biometric trait to bypass system
security.
• The biometric data obtained by a person during the
authentication process can vary significantly from the
data used to produce the template during enrollment.
That can have a big effect on the matching process.
 Multi-Biometric Systems
• One way to improve the accuracy of biometric
systems is to use more than one biometric trait in a
recognition application.
• For example, the face and iris traits, or the
fingerprints from all the ten fingers of an individual
may be used together to resolve the identity of an
individual.
• Such systems are known as multi-biometric systems.
• These systems are expected to be more accurate and
reliable due to the availability of multiple pieces of
evidence
 Data Collection
• The collection of biometric data from a subset of the
targeted population.
• This data is required both for designing the feature
extraction and matcher modules as well as for the
evaluation of the designed biometric system.
• Due to the involvement of human subjects, legal and
privacy issues must also be considered and approval
of organizations like the Institutional Review Board
(IRB) is mandatory in many countries.
• This makes biometric data collection a time-
consuming, relatively expensive, and cumbersome
process
 Data collection

Choice of features and matching algorithm

• Another important factor affecting the choice of
features and matching algorithm is the
interoperability between biometric systems
• For eg, it challenging to compare voice samples
originating from two different handset
• The performance of face recognition algorithms is
severely affected when the images used for
comparison are captured using different camera
types.
 Evaluation
• It requires experts from a variety of fields, including
statistics, computer science, engineering, business, and
psychology, as well as system designers and the end user
community
• In order to understand the performance of a biometric
system, one must address the following questions
– What are error rates of the given Biometric System for a given
application?
– What is the reliability, availability and maintainability of the
system?
– What are the vulnerabilities in the system? And so on
 Evaluation
• The evaluation requires an independent third party to
design, administer, and analyze the test.
• We can divide the matching performance evaluation
of a biometric system into three stages
1. Technology evaluation – Compares different
algorithms like Fingerprint Verification Competitions
(FVC), the Fingerprint Vendor Technology Evaluation
(FpVTE), the Face Recognition Vendor Tests (FRVT),
the Face Recognition Technology (FERET) program,
and the NIST Speaker Recognition Evaluations (SRE)
• These algorithms are examples of biometric
technology evaluations.
 Evaluation
2. Scenario evaluation - the testing of the prototype
biometric systems is carried out in an environment
that closely resembles the real-world application
3. Operational evaluation - is used to ascertain the
performance of a complete biometric system in a
specific real-world application environment on a
specific target population.
 Applications of Biometric
Systems
Commercial applications such as
• computer network login,
• electronic data security,
• e-commerce, Internet access,
• ATM or credit card use,
• physical access control,
• mobile phone, PDA,
• health record management,
• distance learning, etc.
 Applications of Biometric
Systems
Government applications such as
• national ID card,
• managing inmates in a correctional facility,
• driver’s license,
• social security,
• welfare-disbursement,
• Border control,
• passport control, etc
 Applications of Biometric
Systems
Forensic applications such as
• Corpse identification,
• Criminal investigation,
• Missing children,
• parenthood determination, etc.
 Session 3 and 4
• Introduction to Uni-model System
• Introduction to Multi-model Biometric System

• Biometric System Errors

• Performance Measures
 Categories of Biometrics Systems

• Biometric System are basically categorized as

– Unimodal
– Multimodal
 Uni-modal Systems

• Biometric systems that operate using any

single biometric characteristic
• Less expensive and simple
• Unimodal Systems
– Face
– Fingerprint
– Iris
– Ear
– Signature
– Gait
 Unimodal
Limitations of Unimodal Biometric System:
• Susceptibility of biometric sensor to noisy or bad
data.
• The captured biometric trait might be distorted
due to imperfect acquisition conditions.
• (Eg: In fingerprint recognition where a scanner is
unable to read dirty fingerprints clearly and leads
to false database match.)
 Limitations

• It might not be compatible with certain groups of

population
• (fingerprint images might not be properly captured for the
elderly and young children because of faded fingerprints or
underdeveloped fingerprint ridges.)
• Within a large population ,uni-modal biometrics is prone to
inner class similarities
• (eg: facial recognition may not work correctly for identical
twins as the camera might not be able to distinguish
between two subjects leading to inaccurate matching.)
 Limitations
• Unimodal biometrics systems are vulnerable to
spoof attacks where the data can be imitated or
forged.
• (fingerprint recognition systems can be easily
spoofed using rubber fingerprints)
 In a Nutshell, the Limitations are
• Noisy Data
– Leads to false rejection
• Inter-class similarity
– For Identical twins face recognition system will not work
• Incompatible for subset of population
– Hard workers have poor fingerprint pattern
• Vulnerable to spoofing
– Data can be imitated or forged
– Example: Latent fingerprints
• Accuracy
 Multimodal Biometric System

• A biometric system that uses more than one

– Classifier/ Algorithm
– Sample
– Sensor
– Trait
 Need of Multimodal Biometrics

• Reduces error rates of Unimodal

System
• A secondary means of authentication
– Poor quality sample from the sensor
– Non-availability of data
• Combat spoof attacks such as fake fingers
Multimodal Biometric System
 Multimodal Biometric System
• Merge two or more biometric technologies
such as facial recognition ,fingerprint, iris
scanning, hand geometry, voice recognition etc.
• Systems take input from single or multiple
sensors for measuring two or more different
biometric characteristics.
• Improving recognition rate, combining two or
more biometric modalities might be more
appropriate for different applications.
 Integration Scenarios

• Multiple Sensors
• Multiple Biometrics
• Multiple units of same biometrics
• Multiple snapshots of same biometrics
• Multiple Classifiers
 Types of Multimodal Biometric
System
• Multi algorithmic biometric system: System
take a single sensor and then process it using
two or more different algorithms.
• Multi instance biometric systems: Systems
use one or more sensors to capture samples
of two or more different samples of same
biometric trait.(Capturing images of multiple
fingers)
Multi Sensorial Biometric Systems
• Multi sensorial biometric systems: system use
two or more distinctly different sensors to
capture the same instance of a biometric trait.
• Captured samples are then processes using
single algorithm or combination of algorithm
• ( eg: same facial image is captured using
visible light camera and infrared camera fixed
with particular frequency)
 Fusion Strategies

• Multimodal biometric system requires

integration of data of different modalities
like face, fingerprint, retina, voice, iris, etc
• It can be done through a process called
“Fusion”
• There are different fusion methods or
strategies used in the mult-modal Biometric
Systems
 Different Fusion Strategies

• Fusion prior to matching

– Sensor level fusion
– Feature level fusion

• Fusion after matching

– Match score fusion
– Rank level fusion
– Decision level fusion
 Sensor Level Fusion

Sensor level fusion

• We fuse the biometric traits coming from the different
sensors such as fingerprint scanner, iris scanner, video
camera etc. to form a merged biometric trait and process.
– Raw data from the sensor(s) are combined.
– This is referred to as image level or pixel level fusion.
– Sensor level fusion can benefit multi-sample systems which
capture multiple snapshots of the same biometrics.
– For example, 2D face images of an individual obtained from
several cameras can be combined to form a 3D model of the
face.
 Feature Level Fusion

• Feature level fusion: signals coming from different

biometric channels are first processed after which
the feature vectors are extracted separately from
each biometric trait.
– It refers to combine different feature sets extracted from
multiple biometric sources.
– When feature sets are homogeneous, a single resultant feature
vector can be calculated as a weighted average of the individual
feature vector
– When the feature set are non-homogeneous , we can
concatenate to form a single feature vector.

 Match Score Fusion

• Match score fusion: individual matching score is found,

we then fuse the matching level to find composite
matching score which will be used for classification.
– Scores generated from different matching modules are combined to
produce a single score.
– Final decision is taken by considering the fused score.
– Normalization and Similarity/ Dissimilarity Score
– There are various approaches possible for combining the individual
scores.
 Product rule
 Sum rule
 Weighed sum rule
 Max rule and median rule
 Rank Level Fusion

Rank level fusion: For identification, output is the

ranks of enrolled identities.This fusion scheme is to
consolidate the ranks of individual biometric
systems to derive a fused rank for each identity.
– It reveals less information than match scores. However,
unlike match scores, the ranking output by multiple
biometric systems are comparable.
– No normalization is needed and this makes the rank
level fusion schemes simpler to implement compared to
the score level fusion techniques.
 Highest rank method
 Logistic regression method
 Decision Level Fusion

Decision level fusion: Each biometric trait is first pre

classified separately. Individual trait is first captured and
then features are extracted from the captured trait.
 Decision level fusion is the highest level fusion of biometric evidences.
 Fusion is carried out at the abstract or decision level when only the decisions
output by the individual biometric matchers are available.
 It logically combines accept/reject matching decisions of different matchers.
 “AND” and “OR” rule
 Majority voting
 Weighted majority voting
 Bayesian decision fusion
 Advantages of Multimodal Biometric
Systems
• Reduce data distortion.
• Very difficult to spoof
• Robust, reliable and accurate as compared to
unimodal systems.
• Overcome the various limitations of unimodal
systems and hence are suitable to many
industries such as healthcare, civil id and
financial industries.
 Biometric System Errors
• Uniqueness and permanence of the underlying biometric
trait are the fundamental premises of a Biometric System
• Biometric identifier is said to be unique only if any two
persons in the world can be differentiated based on the
given identifier.
• A biometric trait is permanent if it does not change over
the lifetime of an individual.
• Biometric systems rely only on the digital measurements
of the body characteristics, and not the real physical
traits.
• This process of measurement (sensing) introduces
variations in the samples of the same biometric trait of a
 Biometric System Errors
• The variability observed in the biometric feature set of an
individual is known as intra-user variations or intra-class
variations.
• This variability may be due to reasons like
– imperfect sensing conditions (e.g., noisy fingerprint due to
sensor malfunction),
– alterations in the user’s biometric characteristics, changes in
ambiant conditions (e.g., inconsistant illumination levels in face
recognition applications), and
– variations in the user’s interaction with the sensor (e.g.,
occluded iris)
• An ideal biometric feature set must exhibit small inter-
user similarity and small intra-user variations
 Biometric System Errors
• Two types of errors, namely,
– false non-match and
– false match
• When the intra-user variation is large, two samples of the
same biometric trait of an individual (mate samples) may
not be recognized as a match, and this leads to a false
non-match error.
• A false match occurs when two samples from different
individuals (non-mate samples) are incorrectly
recognized as a match due to large inter-user similarity.
 Performance measures
• The basic measures of the accuracy of a biometric system are
– False Non-Match Rate (FNMR) and
– False Match Rate (FMR).
• FNMR refers to the expected probability that two mate
samples (samples of the same biometric trait obtained from
the same user) will be falsely declared as a non-match.
• FMR is the expected probability that two non-mate samples
will be incorrectly recognized as a match
• A FNMR of 5% indicates that on average, 5 in 100
authentication attempts by genuine users will not succeed.
• A FMR of 0.02% indicates that on average, 1 in 5,000
authentication attempts by random impostors are likely to
succeed
 Performance measures
Verification system error rates
• In the context of biometric verification, FNMR and FMR are
generally referred to as False Reject Rate (FRR) and False
Accept Rate (FAR), respectively
• A match score is termed as a genuine or authentic score if
it indicates the similarity between two mate samples.
• An impostor score measures the similarity between two
non-mate samples
• Given a set of genuine and impostor match scores, FRR can
be defined as the proportion of genuine scores that are less
than the threshold η and
• FAR can be defined as the fraction of impostor scores that
are greater than or equal to η.
 Performance measures
• We will use the labels ω0 and ω1 to denote the impostor and
genuine classes, respectively. Let p(s|ω1) and p(s|ω0) be the
probability density functions of the genuine and impostor scores,
respectively
• The FAR and FRR of the biometric system are given by

• If the threshold is increased, FAR will decrease but the FRR will
increase and vice versa. Hence, for a given biometric system, it is
not possible to decrease both these errors simultaneously by
varying the threshold.
• The Genuine Accept Rate (GAR) or True Accept Rate (TAR) can be
used as an alternative to FRR
 Four Categories of Users
• Four categories of users are usually defined in the
biometrics (Doddington’s zoo)
– Sheep represent users whose biometric feature sets are very
distinctive and exhibit low intra-class variations. Therefore,
these users are expected to have low false accept and false
reject errors.
– Goats refer to users who are prone to false rejects. The
biometric feature sets of such users typically exhibit large intra-
class variations
 Performance measures
– Lambs are users whose biometric feature set overlaps
extensively with those of other individuals. The biometric
feature sets of these users have high inter-user similarity. Thus,
a randomly chosen user (from the target population) has a
higher probability of being accepted as a lamb than as a sheep.
The false accept rate associated with these users is typically high
– Wolves indicate individuals who are successful in deliberately
manipulating their biometric trait (especially behavioural traits)
in order to impersonate legitimately enrolled users of a system.
Since the wolves make a concerted effort to adopt the identity
of another user, such an effort is often referred to as an
adversary attack and it can increase the FAR of a system.
Examples include forging the signature of another user or
mimicking someone else’s voice
 Performance measures
• The Failure to Enroll (FTE) rate denotes the proportion of
users that cannot be successfully enrolled in a biometric
system
• This necessitates the design of robust and efficient user
interfaces that can assist an individual both during
enrollment and recognition
• The fraction of authentication attempts in which the
biometric sensor cannot capture the sample presented to
it is known as Failure to Capture (FTC) or Failure to
Acquire (FTA) rate
• Thus, periodic sensor maintenance is instrumental for the
efficient functioning of a biometric system
 Performance measures
• The performance of a biometric system may also be
summarized using other single-valued measures such as
the Equal Error Rate (EER) and the d-prime value
• The EER refers to that point in a Detection Error Tradeoff-
DET (or) Receiver Operating Characteristic - ROC curve
where the FAR equals the FRR; a lower EER value,
therefore, indicates better performance.
• The d-prime value (d) measures the separation between
the means of the genuine and impostor probability
distributions in standard deviation units and is defined as
 Performance measures
• where μ1 (μ0) and σ1 (σ0) are the mean and standard
deviation, respectively, of the genuine (impostor) score
distributions.
• A higher d-prime value indicates better performance.
• Another single valued performance measure is known as
the F-Ratio, which is defined as

• If the genuine and impostor distributions are Gaussian,

then the EER and F-ratio are related according to the
following expression where
 Performance measures
• Identification system error rates
• Suppose that a biometric identification system, with N
identities enrolled, outputs a set of identities
corresponding to the top t matches (1 ≤ t N).
• The identification rank is defined as the rank of a user’s
correct identity in the top t matches returned by the
identification system.
• There are two types of identification system errors. A
false positive identification occurs when an identity is
returned for a user not enrolled in the system.
• This is analogous to the false match case in biometric
verification
 Performance measures
• The expected proportion of identification transactions by
users not enrolled in the system, where an identity is
returned, is known as the false positive identification rate
(FPIR).
• The FPIR depends both on the size of the enrollment database
(N) and the threshold (η)
• False negative identification refers to a scenario where the
transacting user is enrolled in the database, but his correct
identity is not among those returned by the system.
• A quantity related to the FNIR is the true positive
identification rate (TPIR), which is the expected proportion of
identification transactions by users enrolled in the system,
where the user’s correct identity is among the t identities
returned by the system. Therefore, FNIR = 1− TPIR
End of Part 1 (UNIT 1)