Face Recognition using PCA and Eigen Face Approach
1 like85 views
This document discusses a face recognition system based on principal component analysis (PCA) and the eigenface approach. The system consists of two levels of authentication: face recognition and password verification via SMS. In the PCA implementation, training images are used to calculate eigenfaces, which are then used to extract features from input images. Euclidean distance between the input and training image features is calculated to classify faces as known or unknown. The system was implemented in MATLAB and experiments showed it could distinguish individual faces from a database. A GSM modem was also interfaced to send one-time passwords to users' phones for additional authentication.
![International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 04 Issue: 12 | Dec-2017 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 460
Face Recognition using PCA and Eigen Face Approach
Harshada Ashok Kardile
M.E (VLSI and Embedded Systems), G. H. Raisoni Institute of Technology
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - This paper mainly presents face recognition
system based on principal component analysis. The goal is
to implement the system which is able to distinguish a
single face from the larger database. System consists of
two levels of authentication first is the face recognition
stage and second is the password verification stage.
Password is send to person’s mobile number through GSM.
Proposed algorithm is implemented using MATLAB
software. A number of experiments were done to evaluate
the performance o the system.
Key Words: Face Recognition, Principle Component
Analysis, Eign Values, GSM Modem, Person
Authentication,Verification.
1. INTRODUCTION
Over the last decade, face recognition has become
important research area in computer vision and it has
wide range of applications such as Information Security,
Access Management, Biometrics, Law Enforcement,
Personal Security, and Entertainment –Leisure. Face
recognition could be done under variety of conditions like
front view, 45 degree view, faces with spectacles, scaled
front view [1].
In any face recognition system, feature extraction and
classification are two fundamental stages. It is necessary
to enhance these stages for improving performance of the
ace recognition system. Feature extraction reduces the
dimensionality of the image by using linear or non-linear
transformations of the images with its successive selected
feature, so that representation of exacted feature is
possible. There are variations with feature extraction of
human face as there are some problems such as various
backgrounds, aging, illumination, and lightning condition.
In this paper PCA is used for identification and pattern
recognition [2].
This paper discusses about the face recognition process
in section 2. Proposed system block diagram and system
flow is given in section 3 Principal Component Analysis
and its implementation given in section 4. Results of the
system are given in section 5. Finally the work is
concluded in section 6.
2. FACE RECOGNITION SYSTEM
One of the easy and effective PCA approaches used in face
recognition systems is eigen face approach. In this
approach faces are transformed into a small set of
essential characteristics, called as eigen faces. Eigen faces
are the main components of the training set. In
classification process, a new image is projected in the
Eigen face subspace, and then the person is classified by
comparing its position in eigen face space with the
position of known individuals.
Recognition process involves two stages
1. Initialization Process
The Initialization process consists of following steps
a. Prepare training set of face images
b. Calculate eigen faces from training set by taking only
highest values These M number of images will define the
face space. As newly added faces are experienced, eigen
faces for these images can be recalculated or updated.
c. Calculate distribution of this M-dimensional space for
the known person by projecting face images
onto the face-space [3].
2. Recognition process
The recognition process consists of following steps
a. Find weight vectors from the input image and also
calculate the M number of eigen faces by projecting the
input image on the Eigen faces.
b. Determine if all the images are of face whether its
known or unknown, by checking if the image is sufficiently
close to a―free space [3].
Fig -1: Block Diagram of the proposed system](https://image.slidesharecdn.com/irjet-v4i1285-180111072933/85/Face-Recognition-using-PCA-and-Eigen-Face-Approach-1-320.jpg)
![International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 04 Issue: 12 | Dec-2017 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 461
c. If it is a face is detected, then classify its weight pattern
as known person or as unknown person.
d. Update the obtained eigen faces or weights as either a
known person or unknown person.
3. PROPOSED SYSTEM
3.1 Block Diagram of the system:
Block diagram of the proposed system is shown in
figure 2.In proposed system we have implemented face
recognition algorithm based on principal component
analysis. First, We have created the data base of the face
images in which extract features of these stored images
which a recognition takes place. We extract color and
shape of the eyes, nose etc. Then, we capture image which
we need to recognize. Perform preprocessing of the image
like noise removal.RGB to Gray conversion of image. Then,
perform feature extraction of the desired face image. Then,
apply Principal Component Analysis which recognize and
gives decision as person is authorized. MATLAB creates
one time password and send it to mobile number of the
registered person. If the person enters valid OTP person
gets authorization. Two level authorizations is done in
proposed face recognition system.
3.1 Modules of the system:
A. Acquisition Module
This is the first part of the face recognition process. The
user gives the input to face recognition system in this
module.
B. Pre-Processing Module
In this face images are normalized to improve
the recognition of the system. The pre-processing steps
consists of following:
Grayscale conversion: It converts the color image into a
gray image. This method is based on different color
transform. According to the R, G, B contents in the image, it
calculates the corresponding value of gray value, and
obtains the gray image at the same time.
C. The Feature Extraction Module
After the pre-processing face image is
given as input to the feature extraction module which
module composes a feature vector and also represent the
face image.
D. The Classification Module
With the help of principal component analysis, the
extracted features are compared with the stored in the
face database. Then face images are classified as known or
unknown [4].
4. SYSTEM IMPLEMENTATION
4.1Principal Component Analysis:
Principal components analysis is a algorithm which
identifies smaller number of uncorrelated variables, these
uncorrelated variables are called as "principal
components", from a large set of data. This method gives
high amount of variance with the less number of principal
components. PCA is mostly for data analysis. In Principal
Component Analysis, we do eigen value decomposition of
a data correlation matrix [5].
Fig -2: Proposed System Flow
Steps in the Principal Component Analysis
1) Get set of database images and then find mean of the
images.
2) Find the difference between mean desired image and
each image in the database.
3) Find covariance matrix of the matrix which we obtained
in step 2.
4) Find Eigen vectors and Eigen values then find the Eigen
faces which has larger eigen values.
5)From eigen Eigen faces find weight vector
6) Similarly find the weight vector for the desired image
which is to be tested.
7) Measure Euclidian distance between weight vectors of
desired image and database images.
8) If the euclidean distance is less than threshold defined
then desired test image is considered as authorized image
[6].](https://image.slidesharecdn.com/irjet-v4i1285-180111072933/85/Face-Recognition-using-PCA-and-Eigen-Face-Approach-2-320.jpg)
![International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 04 Issue: 12 | Dec-2017 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 462
Fig -3: Flow diagram of the principal component analysis
5. IMPLEMENTATION RESULTS
5.1 Image Pre-processing Results:
Input face image is the RGB image we convert RGB image
to grayscale image.RGB to grayscale image is shown in
figure 5.
Fig -4: RGB Input Image Fig -5: Grayscale Input Image
5.2 Training Set:
Fig -6: Training Set
5.3 PCA Results:
1. Mean Face Image
Fig -7: Mean Face Image
Fig -8: Eigenface ranked according to usefulness
6. CONCLUSION
This paper presents the face recognition system
based on principal component analysis and eigen face
approach. Proposed algorithm is implemented using
MATLAB software.GSM is also interfaced to send
password on person’s mobile number which allows two
levels of person authentication. Advantages of using PCA
are data can be compressed without loss. This approach is
preferred due to its speed, simplicity, learning capability.
Further, this system can be extended to recognize the
gender of a person or to interpret the facial expression of a
person.
REFERENCES
[1] Liton Chandra Paul , Abdulla A Sumam ,“Face
Recognition Using Principal Component Analysis
Method”, International Journal of Advanced Research
in Computer Engineering & Technology
(IJARCET),Volume 1, Issue 9, November 2012.](https://image.slidesharecdn.com/irjet-v4i1285-180111072933/85/Face-Recognition-using-PCA-and-Eigen-Face-Approach-3-320.jpg)
![International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 04 Issue: 12 | Dec-2017 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 463
[2] Kiran D. Kadam, “Face Recognition using Principal
Component Analysis with DCT”, International Journal
of Engineering Research and General Science Volume
2, Issue 4, June-July, 2014.
[3] S. Thakur1, J. K. Sing*2, D. K. Basu2, M. Nasipuri2, M.
Kundu “Face Recognition using Principal Component
Analysis and RBF Neural Networks”, IJSSST, Vol. 10,
No. 5.
[4] Ningthoujam Sunita Devi1, K. Hemachandran2 ,“Face
Recognition Using Principal Component
Analysis”, Ningthoujam Sunita Devi et al, / (IJCSIT)
International Journal of Computer Science and
Information Technologies, Vol. 5 (5) , 2014, 6491-
6496
[5 ] Madhuri M. Ghodake, Parul S. Arora, “Face Recognition
Using Principal Component Analysis for Security
Based System”, International Journal of Science and
Research (IJSR) ,ISSN (Online): 2319-7064.
[6] Saurabh P.Bahurupi, D.S.Chaudhari ,“Principal
Component Analysis for Face Recognition”,
International Journal of Engineering and Advanced
Technology (IJEAT),ISSN: 2249 – 8958, Volume-1,
Issue-5, June 2012.
[7] K.Swathi, T.Siva “Face recognition technique for door
phone embedded system”, IOSR Journal of Electronics
and Communication Engineering (IOSR-JECE)e-ISSN:
2278-2834,p- ISSN: 2278-8735.Volume 8, Issue 4
(Nov. - Dec. 2013), PP 19-22.
[8] Vytautas PERLIBAKAS ,“Face Recognition Using
Principal ComponentAnalysis and Wavelet Packet
Decomposition”, INFORMATICA, 2004, Vol. 15, No. 2,
243–250 243,2004Institute of Mathematics and
Informatics, Vilnius.
[9] Mrs. Sunita Roy1 and Mr. Susanta Podder “Face
detection and its applications”, IJREAT International
Journal of Research in Engineering & Advanced
Technology, Volume 1, Issue 2, April-May, 2013.
[10] Prof.Joshi Shweta, Mr. Mahadik Vaibhav, Mr. Thorat
Vishal, Mr. Pachupate Rushikesh ,“Automatic Face
Recognition System”, International Journal of
Advanced Research in Electronics and Communication
Engineering (IJARECE) Volume 5, Issue 2, February
2016.
[11] J meghana, 2kavyashree k, 3naveen kumar g, 4
supriya h s, 5c gururaj “iris detection based on
principal component analysis with gsm interface”,
oceedings of IRF International Conference, 29th &
30th October, 2016, Chennai, India, ISBN: 978-93-
82702-18-4.](https://image.slidesharecdn.com/irjet-v4i1285-180111072933/85/Face-Recognition-using-PCA-and-Eigen-Face-Approach-4-320.jpg)
Face Recognition using PCA and Eigen Face Approach
- 1. International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 04 Issue: 12 | Dec-2017 www.irjet.net p-ISSN: 2395-0072 © 2017, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 460 Face Recognition using PCA and Eigen Face Approach Harshada Ashok Kardile M.E (VLSI and Embedded Systems), G. H. Raisoni Institute of Technology ---------------------------------------------------------------------***--------------------------------------------------------------------- Abstract - This paper mainly presents face recognition system based on principal component analysis. The goal is to implement the system which is able to distinguish a single face from the larger database. System consists of two levels of authentication first is the face recognition stage and second is the password verification stage. Password is send to person’s mobile number through GSM. Proposed algorithm is implemented using MATLAB software. A number of experiments were done to evaluate the performance o the system. Key Words: Face Recognition, Principle Component Analysis, Eign Values, GSM Modem, Person Authentication,Verification. 1. INTRODUCTION Over the last decade, face recognition has become important research area in computer vision and it has wide range of applications such as Information Security, Access Management, Biometrics, Law Enforcement, Personal Security, and Entertainment –Leisure. Face recognition could be done under variety of conditions like front view, 45 degree view, faces with spectacles, scaled front view [1]. In any face recognition system, feature extraction and classification are two fundamental stages. It is necessary to enhance these stages for improving performance of the ace recognition system. Feature extraction reduces the dimensionality of the image by using linear or non-linear transformations of the images with its successive selected feature, so that representation of exacted feature is possible. There are variations with feature extraction of human face as there are some problems such as various backgrounds, aging, illumination, and lightning condition. In this paper PCA is used for identification and pattern recognition [2]. This paper discusses about the face recognition process in section 2. Proposed system block diagram and system flow is given in section 3 Principal Component Analysis and its implementation given in section 4. Results of the system are given in section 5. Finally the work is concluded in section 6. 2. FACE RECOGNITION SYSTEM One of the easy and effective PCA approaches used in face recognition systems is eigen face approach. In this approach faces are transformed into a small set of essential characteristics, called as eigen faces. Eigen faces are the main components of the training set. In classification process, a new image is projected in the Eigen face subspace, and then the person is classified by comparing its position in eigen face space with the position of known individuals. Recognition process involves two stages 1. Initialization Process The Initialization process consists of following steps a. Prepare training set of face images b. Calculate eigen faces from training set by taking only highest values These M number of images will define the face space. As newly added faces are experienced, eigen faces for these images can be recalculated or updated. c. Calculate distribution of this M-dimensional space for the known person by projecting face images onto the face-space [3]. 2. Recognition process The recognition process consists of following steps a. Find weight vectors from the input image and also calculate the M number of eigen faces by projecting the input image on the Eigen faces. b. Determine if all the images are of face whether its known or unknown, by checking if the image is sufficiently close to a―free space [3]. Fig -1: Block Diagram of the proposed system
- 2. International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 04 Issue: 12 | Dec-2017 www.irjet.net p-ISSN: 2395-0072 © 2017, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 461 c. If it is a face is detected, then classify its weight pattern as known person or as unknown person. d. Update the obtained eigen faces or weights as either a known person or unknown person. 3. PROPOSED SYSTEM 3.1 Block Diagram of the system: Block diagram of the proposed system is shown in figure 2.In proposed system we have implemented face recognition algorithm based on principal component analysis. First, We have created the data base of the face images in which extract features of these stored images which a recognition takes place. We extract color and shape of the eyes, nose etc. Then, we capture image which we need to recognize. Perform preprocessing of the image like noise removal.RGB to Gray conversion of image. Then, perform feature extraction of the desired face image. Then, apply Principal Component Analysis which recognize and gives decision as person is authorized. MATLAB creates one time password and send it to mobile number of the registered person. If the person enters valid OTP person gets authorization. Two level authorizations is done in proposed face recognition system. 3.1 Modules of the system: A. Acquisition Module This is the first part of the face recognition process. The user gives the input to face recognition system in this module. B. Pre-Processing Module In this face images are normalized to improve the recognition of the system. The pre-processing steps consists of following: Grayscale conversion: It converts the color image into a gray image. This method is based on different color transform. According to the R, G, B contents in the image, it calculates the corresponding value of gray value, and obtains the gray image at the same time. C. The Feature Extraction Module After the pre-processing face image is given as input to the feature extraction module which module composes a feature vector and also represent the face image. D. The Classification Module With the help of principal component analysis, the extracted features are compared with the stored in the face database. Then face images are classified as known or unknown [4]. 4. SYSTEM IMPLEMENTATION 4.1Principal Component Analysis: Principal components analysis is a algorithm which identifies smaller number of uncorrelated variables, these uncorrelated variables are called as "principal components", from a large set of data. This method gives high amount of variance with the less number of principal components. PCA is mostly for data analysis. In Principal Component Analysis, we do eigen value decomposition of a data correlation matrix [5]. Fig -2: Proposed System Flow Steps in the Principal Component Analysis 1) Get set of database images and then find mean of the images. 2) Find the difference between mean desired image and each image in the database. 3) Find covariance matrix of the matrix which we obtained in step 2. 4) Find Eigen vectors and Eigen values then find the Eigen faces which has larger eigen values. 5)From eigen Eigen faces find weight vector 6) Similarly find the weight vector for the desired image which is to be tested. 7) Measure Euclidian distance between weight vectors of desired image and database images. 8) If the euclidean distance is less than threshold defined then desired test image is considered as authorized image [6].
- 3. International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 04 Issue: 12 | Dec-2017 www.irjet.net p-ISSN: 2395-0072 © 2017, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 462 Fig -3: Flow diagram of the principal component analysis 5. IMPLEMENTATION RESULTS 5.1 Image Pre-processing Results: Input face image is the RGB image we convert RGB image to grayscale image.RGB to grayscale image is shown in figure 5. Fig -4: RGB Input Image Fig -5: Grayscale Input Image 5.2 Training Set: Fig -6: Training Set 5.3 PCA Results: 1. Mean Face Image Fig -7: Mean Face Image Fig -8: Eigenface ranked according to usefulness 6. CONCLUSION This paper presents the face recognition system based on principal component analysis and eigen face approach. Proposed algorithm is implemented using MATLAB software.GSM is also interfaced to send password on person’s mobile number which allows two levels of person authentication. Advantages of using PCA are data can be compressed without loss. This approach is preferred due to its speed, simplicity, learning capability. Further, this system can be extended to recognize the gender of a person or to interpret the facial expression of a person. REFERENCES [1] Liton Chandra Paul , Abdulla A Sumam ,“Face Recognition Using Principal Component Analysis Method”, International Journal of Advanced Research in Computer Engineering & Technology (IJARCET),Volume 1, Issue 9, November 2012.
- 4. International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 04 Issue: 12 | Dec-2017 www.irjet.net p-ISSN: 2395-0072 © 2017, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 463 [2] Kiran D. Kadam, “Face Recognition using Principal Component Analysis with DCT”, International Journal of Engineering Research and General Science Volume 2, Issue 4, June-July, 2014. [3] S. Thakur1, J. K. Sing*2, D. K. Basu2, M. Nasipuri2, M. Kundu “Face Recognition using Principal Component Analysis and RBF Neural Networks”, IJSSST, Vol. 10, No. 5. [4] Ningthoujam Sunita Devi1, K. Hemachandran2 ,“Face Recognition Using Principal Component Analysis”, Ningthoujam Sunita Devi et al, / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol. 5 (5) , 2014, 6491- 6496 [5 ] Madhuri M. Ghodake, Parul S. Arora, “Face Recognition Using Principal Component Analysis for Security Based System”, International Journal of Science and Research (IJSR) ,ISSN (Online): 2319-7064. [6] Saurabh P.Bahurupi, D.S.Chaudhari ,“Principal Component Analysis for Face Recognition”, International Journal of Engineering and Advanced Technology (IJEAT),ISSN: 2249 – 8958, Volume-1, Issue-5, June 2012. [7] K.Swathi, T.Siva “Face recognition technique for door phone embedded system”, IOSR Journal of Electronics and Communication Engineering (IOSR-JECE)e-ISSN: 2278-2834,p- ISSN: 2278-8735.Volume 8, Issue 4 (Nov. - Dec. 2013), PP 19-22. [8] Vytautas PERLIBAKAS ,“Face Recognition Using Principal ComponentAnalysis and Wavelet Packet Decomposition”, INFORMATICA, 2004, Vol. 15, No. 2, 243–250 243,2004Institute of Mathematics and Informatics, Vilnius. [9] Mrs. Sunita Roy1 and Mr. Susanta Podder “Face detection and its applications”, IJREAT International Journal of Research in Engineering & Advanced Technology, Volume 1, Issue 2, April-May, 2013. [10] Prof.Joshi Shweta, Mr. Mahadik Vaibhav, Mr. Thorat Vishal, Mr. Pachupate Rushikesh ,“Automatic Face Recognition System”, International Journal of Advanced Research in Electronics and Communication Engineering (IJARECE) Volume 5, Issue 2, February 2016. [11] J meghana, 2kavyashree k, 3naveen kumar g, 4 supriya h s, 5c gururaj “iris detection based on principal component analysis with gsm interface”, oceedings of IRF International Conference, 29th & 30th October, 2016, Chennai, India, ISBN: 978-93- 82702-18-4.