An Efficient Approach Based On LTP and SVM For 3D Face Recognition
An Efficient Approach Based On LTP and SVM For 3D Face Recognition
An Efficient Approach Based On LTP and SVM For 3D Face Recognition
Instructor
Asst. Prof. Dr. Kamil YURTKAN
Prepared By
Gheyath M. Othman
21911314
2020-2021
Introduction
In the past years, a revolution took place in the world of technology and developed
rapidly in all areas, covering various aspects of life. One of the hottest topics that
researchers works in is computer vision including artificial intelligence. As it has a
great importance, it represents the basics for many applications that are currently
utilized in various sectors. The technology of biometric recognition has progressively
developed specially in security for identification purposes. Such technology is face
recognition, which uses facial information of humans to recognize people.
Introduction Cont..
In this paper a 3D (three-dimensional) face recognition approach is proposed using
LTP (Local Ternary Pattern) with SVM (Support Vector Machine) techniques. LTP
technique which is a variant and extension of LBP (Local Binary Pattern) will be used
for feature information extraction from the 3D faces, and then the SVM technique is
employed to classify and recognize the faces according to extracted features. The
database that depended in this work is Texas 3DFRD (Three-Dimensional Face
Recognition Database), which can be downloaded from the link https://www.face-
rec.org/databases/.
Local Ternary Pattern (LTP)
LTP is an extension of LBP which used for feature extraction. Basically, it capture the
variations of intensity in local neighborhood. Its threshold the pixels in to three values
by using a threshold constant value. If we assume F as a constant threshold, P for
neighboring pixel, and C for the center pixel value, the threshold result is:
1, if p > c + f
0, if p > c – f and p < c + f
-1, if p < c - f
Here, one of the three values for each of the threshold pixel. After thresholding in to
ternary pattern neighbor pixels are merged. A large range will be the result of
computing the histogram of the ternary values, hence ternary pattern is separated in to
two binary patterns. In this case, histograms then concatenated to generate a
descriptor which double the size of LBP.
Local Ternary Pattern (LTP)
Example: 3*3 window, and threshold constant +5 and -5
t=5[45,55]
Training Phase
• In [4], Gabor and Local Ternary Pattern (LTP) features are depended for a 3D face
recognition system, by selecting the significant features by means of GIR and the
classifier ELM (Extreme Learning Machine) is used to classify human faces.
Literature Review
• However, in [5] a Dynamic Local Ternary Pattern descriptor has been proposed
which employed for feature extraction in face recognition, it’s used not only for
encoding significant texture features but it’s also reduces the noise effect in face
recognition. This helped to improve the accuracy.
• In [6], a 3D face recognition study has been proposed based on DCNN. At first, the
images are preprocessed then redundant information are discarded. The features
are extracted from both the 2D and the depth images and followed by the process
of fusion. The study employed SoftMax classifier for recognizing faces.
• In this work the Local Ternary Pattern (LTP) technique will be employed to extract
the feature information from the 3D faces and Support Vector Machine technique
is used to classify and recognize human faces. The performance metrics will be the
accuracy of the proposed approach. The proposed model will be implemented
using Python.
References
1. Agarwal, M., Singhal, A., & Lall, B. (2018). 3D local ternary co-occurrence
patterns for natural, texture, face and bio medical image retrieval.
Neurocomputing, 313, 333–345.
2. Gupta, S., Castleman, K. R., Markey, M. K., & Bovik, A. C. (2010). Texas 3D
Face Recognition Database. 2010 IEEE Southwest Symposium on Image Analysis
Interpretation (SSIAI), 97–100.
3. Shi, L., Wang, X., & Shen, Y. (2020). Research on 3D face recognition method
based on LBP and SVM. Optik- International Journal for Light and Electron
Optics, 220, 165157.
4. M. Chandrakala, S. Ravi, (2018). Effective 3D Face Recognition Technique Based
on Gabor and LTP Features, International Journal of Engineering and Advanced
Technology 2249 – 8958.
References
5. Ibrahim, M., Efat, M. I. A., Shamol, H. K., Khaled, S. M., Shoyaib, M., &
Abdullah-Al-Wadud, M. (2014, January). Dynamic local ternary pattern for face
recognition and verification. In International Conference on Computer
Engineering and Applications (Vol. 1012, No. 8).
6. Jianying Feng, Qian Guo, Yudong Guan, Mengdie Wu, Xingrui Zhang, Chunli Ti,
(2018). 3D Face Recognition Method Based on Deep Convolutional Neural
Network, Smart Innovations in Communication and Computational Sciences,
pp.123-130.
Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.
Alternative Proxies: