Face Granulation Using Difference of Gaussian (DOG) Method For Face Recognition

International Journal of Advance Foundation and Research in Computer (IJAFRC)
Volume 2, Issue 6, June - 2015. ISSN 2348 4853
Face Granulation using Difference of Gaussian (DOG)

Method for Face Recognition
Asavari G. Joshi* PG Student, Prof. A. S. Deshpande, Assistant Professor
E & TC Department JSPMs, Imperial College of Engineering & Research Wagholi, Pune, India
asavarigjoshi@gmail.com *, panuradha2010@gmail.com
ABSTRACT
In this paper first we represent an overview of face recognition and discuss the methodology and
functioning. This paper presents surgically altered robust face recognition based on granular
computation and features extraction. Biometrics for person authentication has instigated several
techniques for evading identification. The results which are obtained from this identification can
be used for recognition of face images. In this paper granular levels are computed. The Gaussian
operator generates a sequence of low pass filtered images. Then, DOG pyramid is formed and by
this granulation, facial features are segregated at different resolutions to provide noise, edge
information smoothness present in a face image. . In this paper analysis up to Face Granulation is
done and the results are explained. The MATLAB software is used for this technique.
Index Terms: Face recognition, Pre-processing, Face detection, Face granulation.
I. INTRODUCTION
Face recognition becomes a most important biometrics authentication technique from the past few years.
Face recognition is a biometric approach to verify or recognize the identity of a living person based on
his/her physiological characteristics. Biometrics can measure both physiological and behavioural
charactrics. In behavioural biometric, data derived from an action. In physiological biometric, data
derived from direct measurement. Both these biometrics are based on measurements. Biometric is the
technology is motivated by traditional methods. Traditional methods can be categorized as possessionbased methods and knowledge-based methods. In possession-based method, items such as keys, badges,
or cards are utilized. This method requires low cost. However, these items can be shared, duplicated and
easily lost. In knowledge-based method, there are also drawbacks for those who used password or PIN.
Some passwords are easily guessed. Besides, they can be forgotten or shared. Face Recognition is integral
part of biometrics. Face recognition is a successful application of image analysis and pattern recognition.
The main aim of face recognition is to recognize person from video or pictures using databases of face
images. Basically there are three steps in face recognition system; face detection, feature extraction, and
classification [7].
The face is our primary focus of attention in social intercourse, playing major role in conveying identity
and emotion. Although the ability to infer intelligence or character from facial appearance is suspect,
the human ability to recognize faces is Remarkable. We can recognize thousands of faces learned
throughout our lifetime and identify familiar faces at a glance even after years of separation. This skill is
quite robust, despite large changes in the visual stimulus due to viewing conditions, expression, aging,
and distraction such as glasses, beards or changes in hairstyle. Face recognition has become an
important issue in many applications such as security systems, credit card verification and criminal
identification. Therefore, face recognition is a very high-level computer vision task, in which many
early vision techniques can be involved. Face recognition is also being used in conjunction with other
50 | 2015, IJAFRC All Rights Reserved
www.ijafrc.org

biometrics such as speech, ear, fingerprint, and iris and gait recognition in order to enhance the
recognition performance of these methods [10, 11].
There are three major research groups, which propose three different approaches to the face recognition
problem. The largest groups have dealt with facial characteristics, which are used by human beings in
recognizing Individual faces. The second group performs human face identification based on feature
vectors extracted from profile silhouettes. The third group uses feature vectors extracted from a frontal
view of the face.
There are two modes in which biometric systems can perform: identification and verification.
In face identification, once a user has his/her face sample presented to biometric system, the sample is
compared with every biometric reference stored in database. Face acquisition module is the entry point,
where face images are captured by camera. Face detection locates and segments certain face region from
cluttered scenes. Alternately, several pre-processing techniques can be applied on face images for
enhancement, such as histogram equalization, filtering, background removal, illumination normalization,
size normalization. Afterwards, relevant features are extracted from face images and presented to
classification based on identification concept. Face identification system has to determine the identity of
the input face image based on comparisons among all templates stored in database. It is a one to many (1:
N) application. This application is typically used for surveillances and law enforcement investigations.
Second, in face verification, users face template is compared with specific face biometric references in
the database corresponding to the users identity. While giving biometric sample, the user has to claim
identity by entering name, or presenting ID cards, token. With this claimed identity, biometric system
compares the biometric sample with associated biometric references. Finally the system yields
verification results i.e., true or false. Verification is a one to one (1:1) concept which is commonly applied
in access control applications.
Additionally face recognition methods categorize as,
1) Feature based Approach
In feature-based approaches, geometric features such as position and width of nose, eyes, and mouth
eyebrows thickness, are first extracted to represent a face. Like facial features other fiducial marks are
also extracted. Then compute geometric relationship among those facial points. Appearance based
method give good results than feature based method [1]. Benefits of using feature based schemes include
high speed matching.
2) Holistic Approach
Holistic face recognition utilises global information from faces to perform face recognition. The Global
information from faces is represented by a small number of features which are directly derived from the
pixel information. This small number of features distinctly captures the variance among different
individual faces [2]. In Holistic Approach, whole face is taken as an input to perform face recognition in
face detection system. The main advantage offered by Holistic approach is that they concentrate on only
limited regions or points of interest without destroying any of the information in images. Hybrid
approach is combination of feature based approach and holistic approach. In this, both whole and local
face is used as input to face detection system.
Today security is very important. Face recognition is a challenging problem in the field of pattern
recognition and image processing. There are numerous application areas such as, access control, credit
card verification, surveillance etc. in which face recognition plays very important role. Face recognition
www.ijafrc.org

has received great attention because of lots of applications. Face recognition is challenging task in terms
of hardware that is creating physical implementation and software that is developing algorithmic
solutions [6].
II. LITERATURE REVIEW
This section gives an overview on the major human face recognition techniques. There are different
methods used in face recognition. Every method has its advantages and disadvantages. Some methods
are according to feature based and on similarity based.
Neelam Mahale and Dr. M.S. Nagmode [3] use principal Component Analysis method in to improve the
security of an Automated Teller Machine. They used face recognition in one part to access the Automated
Teller Machine. This paper presents a Face Recognition method using Principal Component Analysis. This
method applies on both data base image and input image. By the use of PCA the system finds the Eigen
values, Eigen vector and Euclidian distance. After comparing from database it declares the matches. PCA
has some drawbacks, like Poor discriminatory power, and High computational load.
A. Gunjan Dashore and B. Dr. V. Cyril Raj [4] proposed an efficient method for face recognition using
Principal Component Analysis. It is one of the most popular representation methods for a face image.
Karhunen-Love is based on the Eigen faces technique in which the Principal Component Analysis (PCA)
is used [8]. The drawback of Eigen face is, it is sensitive for lightening conditions and position of the
Head. Disadvantage is finding the eigenvectors and eigenvalues are time consuming.
Lih-Heng Chan [5] proposed a framework of facial biometric was designed based on two subspace
methods i.e., Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA). The fisher
face method for face recognition described by Belhumeur et al [9] uses both linear discriminant analysis
and principal component analysis which produce a subspace projection matrix. First, PCA is used for
dimension reduction, where original face images are projected into lower-dimensional face
representations. Second, LDA was proposed to provide a solution of better discriminant. Both PCA and
LDA features were presented to Euclidean distance measurement which is conveniently used as a
benchmark. LDA-based methods outperform PCA for both face identification and verification. The
disadvantage of fisher face method is it is more complex than Eigen face to finding the projection of free
space.
There are several researches present on face recognition in the literature. Principal Component Analysis
is most useful and also proper method. In the present paper we propose a Difference of Gaussian (DOG)
method which can help for face recognition. In feature extraction stage WLD descriptor and Gabor filter
bank will be then used for feature extraction. Further matching will be done between input and original
samples using correlation matrices.
III. METHODOLOGY
The procedure for surgically altered face recognition consist the following steps:
1) Pre-processing 2) Face Detection 3) Face Granulation 4) Feature Extraction 5) Feature Matching
Our main aim is to calculate granule levels and then create DOG pyramid. The whole process is shown in
Fig. 1.
1. Pre-processing
www.ijafrc.org

In the present paper pre-processing step converts the image according to the need of the next level. Preprocessing is done to remove the noise and reliance on the precise registration. The pre-processing is
done by converting input image which is in RGB format into Gray format.
Fig. 1 System Overview

2. Face Detection
Face detection is the process to extract face regions from input image which has normalized intensity and
uniform in size and shape. It detects face and ignores anything else. An executable (.dll-dynamic link
library) file is utilized to extract face region.
3. Face Granulation
Granulation is the act or process of forming or crystallizing into grains. This approach is used to
represent the facial information in several parts to extract the features. For detection of face granules, 2D
Gaussian low pass filter is used to generate difference of Gaussian (DOG). Then image will be down
sampled to make difference of Gaussian pyramid at each iteration level.
DOG flow is shown in Fig. 2
www.ijafrc.org

Fig. 2 DOG flow

4. Feature Extraction
In features extraction stage, WLD descriptor represents an image as a histogram of differential
excitations and gradient orientations. The Gabor filter bank will be then used to extract the features from
face regions to discriminate the illumination changes.
5. Feature Matching
The combined features of weber and Gabor faces will be used for feature matching.
IV. RESULTS
Input image is in RGB format, and then it is converted into Gray scale image for further operation. The
obtained results are shown in figure below,
Fig. 3 Input Image
Fig. 4 RGB to Gray
In above results input image and RGB to Gray image is shown in figure 3 and 4.
Face detection is the process to extract face regions from input image which has normalized intensity and
uniform in size.
www.ijafrc.org

Fig. 5 Detected face

Here seven levels of Granulation are computed. Each image show output of each level. These granules are
used to provide facial features such as smoothness, edge details and blurriness.
Fig. 6 DOG Result

The Detected image shown in figure 4. The final DOG output is shown in figure 5. DOG output is used for
the further processing that is for the feature extraction and matching purpose.
V. CONCLUSION
Face recognition technique is a challenging problem in the field of image processing and computer vision.
The face detection module is used here to obtain face images. Then, DOG pyramid is formed from
successive iterations of Gaussian images. By this granulation, facial features are segregated at different
resolutions to provide edge information, blurriness, noise, and smoothness.
VI. REFERENCES
[1]
FERET Data Set in Wiley Periodicals, 2006.
[2]
Stewart Tseng, Comparison of holistic and feature based approaches to face recognition, M.
Thesis, Royal Melbourne Institute of Technology University, Melbourne, Victoria, Australia. 10
July, 2003.
[3]
Neelam Mahale, Dr. Manoj S. Nagmode and Prajakta S. Ghatol, Face Recognition Using Principal
Component Analysis Method, Int. J. Computer Sci., Vol.-2(7), pp. 57-61 July 2014.
www.ijafrc.org

[4]
A. Gunjan Dashore, B. Dr. V.Cyril Raj, An efficient method for face recognition using Principal
Component Analysis, International Journal of Advanced Technology & Engineering Research
(IJATER), Vol. 2, MARCH 2012.
[5]
Lih-Heng Chan, Sh-Hussain Salleh and Chee-Ming Ting, Face Biometrics Based on Principal
Component Analysis and Linear Discriminant Analysis, J. Computer Sci., 6 (7): 693-699, 2010.
[6]
Janarbek Matai, Ali Irturk and Ryan Kastner, Design and Implementation of an FPGA-based RealTime Face Recognition System, In IEEE International Symposium on Field-Programmable
Custom Computing Machines, 2011, pp. 97-100.
[7]
Riddhi Patel, Shruti B. Yagnik, A Literature Survey on Face Recognition Techniques, IJCTT Vol.-5,
NOV 2013.
[8]
A. S. Tolba, A. H. EI-Baz, A. A. EI-Herby, Face Recognition: A Literature Review, World Academy

of Science Engineering and Technology, Vol 2, 2008.
[9]
Hong Duan, Ruohe Yan, Kunhui Lin, Research on Face Recognition Based on PCA, 978-0-76953480-0/08 2008 IEEE.
[10]
T. Choudhry, B. Clarkson, T. Jebara, and A. Pentland, "Multimodal person recognition using

unconstrained audio and video", in Proceedings, International Conference on Audio and VideoBased Person Authentication, 1999, pp.176-181.
[11]
M. M. Rahman, R. Hartley, and S. Ishikawa, "A Passive and Multimodal Biometric System for
Personal Identification", in International Conference on Visualization, Imaging and Image
Processing. Spain, 2005, pp.89-92.
AUTHORS PROFILE
Miss. Asavari Gajanan Joshi received her B.E. degree from Shivaji University, in
2013. This author is persuing M.E. degree in Signal Processing in Pune University.
www.ijafrc.org

Face Granulation Using Difference of Gaussian (DOG) Method For Face Recognition

Uploaded by

Document Information

Original Title

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

Face Granulation Using Difference of Gaussian (DOG) Method For Face Recognition

Uploaded by

Copyright:

Available Formats

International Journal of Advance Foundation and Research in Computer (IJAFRC)

Volume 2, Issue 6, June - 2015. ISSN 2348 4853