See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/221703163

Race Identification from Face Images

Article · September 2011

CITATIONS READS
15 267

2 authors:

Dr.Hlaing Htake Khaung Tin Myint Myint Sein

University of Computer Studies, Hinthada University of Computer Studies, Yangon
55 PUBLICATIONS   130 CITATIONS    144 PUBLICATIONS   214 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

How the role of information systems evolved in Myanmar View project

The image mining in digital image processing View project

All content following this page was uploaded by Dr.Hlaing Htake Khaung Tin on 09 December 2014.

The user has requested enhancement of the downloaded file.

 ACEEE Int. J. on Information Technology, Vol. 01, No. 02, Sep 2011

Race Identification for Face Images

Hlaing Htake Khaung Tin1 and Myint Myint Sein2
1
University of Computer Studies, Yangon, Myanmar
[email protected]
2
University of Computer Studies, Yangon, Myanmar
[email protected]

Abstract—Humans are able to process a face in a variety of system based on a boosted classifier. The same structure is
ways to categorize it by its identity, along with a number of used for demographic information extraction, including gender
other demographic characteristics, including race, gender , and race. Two categories of race are defined, Asian and non-
and age. Experimental results are based on a face database Asian. Again, their system is focused on low resolution
containing subjects. Race and gender also play an important (24824) images with face data weakly aligned. Their reported
role in face-related applications. Experimental results are
accuracy is about 80%. The other-race effect for face
indicated that participants categorized the race of the face
and this categorization drives the perceptual process. A face recognition has been established in numerous human memory
image data set is collected from Internet, and divided into a studies and in meta-analyses of these studies [11],[12],[13].
training dataset and a test dataset. Experimental results based In fact, the other-race effect in humans can be measured in
on a face database containing 250 subjects. The proposed infants as a decrease in their ability to detect differences in
system can also be applied to other image-based classification individual other-race faces as early as three to nine months
tasks. of age [13].

Index Terms— race identification, PCA, face recognition II. PRE-PROCESSING

I. INTRODUCTION The first step of pre-processing is the face region

extraction. Face region extraction means the input face image
Race refers to classifications of humans into relatively is extracted from input image by using cropping tool. The
large and distinct populations or groups often based on input color image is converted to gray image and stored in
factors such as appearance based on heritable phenotypical database for processing. The input image may be current
characteristics or geographic ancestry, but also often scanned image or realities input image. And then enhancing
influenced by and correlated with traits such as culture, race state occurs. The proposed system allows the free size and
and socio-economic status. As a biological term, race denotes format of color image. Enhancing state is included the noise
genetically divergent human populations that can be marked filtering, gray scale converting, and histogram equalization.
by common phenotypic traits. Humans are able to process a Histogram equalization is mapped the input image’s intensity
face in a variety of ways to categorize it by its identity, along values so that the histogram of the resulting image will have
with a number of other demographic characteristics, including an approximately uniform distribution. The histogram of a
race, gender, and age. Over the past few decades, a lot of digital image with gray levels in the range [0, L-1] is a discrete
effort has been devoted in the biological, psychological, and function.
cognitive sciences areas, to discover how the human brain
perceives, represents and remembers faces. Computational
models have also been developed to gain some insight into
this problem. The demographic features, such as race and
gender, are involved in human face identity recognition. where L is the total number of gray levels , rk is the kth gray
Humans are better at recognizing faces of their own race than level, nk is the number of pixels in the image with that gray
faces of other races [3] [4]. Golby et al. Show that same-race level, n is the total number of pixels in the image, and k = 0, 1,
faces elicit more activity in brain regions linked to face 2, . . . , L - 1. p(rk) is given an estimate of the probability of
recognition [5]. They use functional magnetic resonance occurrence of gray level rk. By histogram equalization, the
imaging (fMRI) to examine if the same-race advantage for local contrast of the object in the image is increased, especially
face identification involves the fusiform face area (FFA), which when the usable data of the image is represented by close
is known to be important for face recognition [6]. Compared contrast values. Through this adjustment, the intensity can
to race identification, the gender classification has received be better distributed on the histogram. This allows for areas
more attention [7] [8] [9]. Gutta et al [9] proposed a hybrid of lower local contrast to gain a higher contrast without
classifier based on RBF networks and inductive decision trees affecting the global contrast.
for classification of gender and race origin, using a 64*72
image resolution. They achieved an average accuracy rate of
92% for the ethnic classification part of the task. Experimental
results for gender classification in Moghaddam and Yang [7]
are based on 21*12 image resolution. Shakhnarovich et al
[10] presented a real-time face detection and recognition
35
© 2011 ACEEE
DOI: 01.IJIT.01.02. 214
 ACEEE Int. J. on Information Technology, Vol. 01, No. 02, Sep 2011

This new subspace is normally lower dimensional (M<< M

<< N2). New basis vectors are defined a subspace of face
images called face space. All images of known faces are
projected onto the face space to find sets of weights that
described the contribution of each vector. By comparing a
set of weights for the unknown face to sets of weights of
known faces, the face can be identified. PCA basis vectors
are defined as eigenvectors of the scatter matrix S defined as:

where  is the mean of all images in the training set and xi is

the ith face image represented as a vector i. The eigenvector
associated with the largest eigenvalue is one that reflects the
greatest variance in the image. That is, the smallest eigenvalue
is associated with the eigenvector that finds the least
variance. A facial image can be projected onto M  M 
dimensions by computing

The vectors are also images, so called, eigenimages, or

eigenfaces.They can be viewed as images and indeed look
like faces. Face space forms a cluster in image space and PCA
gives suitable representation.
B. Nearest Neighbor Classification
One of the most popular non-parametric techniques is
the Nearest Neighbor classification (NNC). NNC asymptotic
or infinite sample size error is less than twice of the Bayes
error [15]. NNC gives a trade-off between the distributions of
the training data with a priori probability of the classes
involved [14]. KNN (Kth nearest neighbor classifier) classifier
is easy to compute and very efficient. KNN is very compatible
and obtain less memory storage. So it has good discriminative
power. Also, KNN is very robust to image distortions (e.g.
rotation, illumination). Euclidian distance is determined
whether the input face is near a known face. The problem of
automatic face recognition is a composite task that involves
detection and location of faces in a cluttered background,
Figure 1. PCA on Myanmar and Non-Myanmar datasets. (a)
“average” Myanmar face; (b) top 12 eigenfaces of Myanmar normalization, recognition and verification.
dataset;(c) “average” Non-Myanmar face; (d) top 12 eigenfaces of
Non-Myanmar dataset. IV. CONCLUSIONS

III. FEATURE EXTRACTION This paper has addressed the race identification problem
based on facial images. The Principal Component Analysis
A. Subspace Face Recognition (PCA) based scheme has been developed for the two-class
The Principal Component Analysis (PCA) can do (Myanmar vs. non-Myanmar) race classification task. An
prediction, redundancy removal, feature extraction, data ensemble framework, which integrates the PCA for the input
compression, etc. Because PCA is a classical technique which face images at multiple scales, is proposed to further improve
can do something in the linear domain, applications having the classification performance of the race identification
linear models are suitable. Let us consider the PCA procedure system. Experimental results based on a face database
in a training set of M face images. Let a face image be containing 250 subjects are encouraging. The normalized
represented as a two dimensional N by N array of intensity classification scores can be used as the confidence with
values, or a vector of dimension N2. Then PCA tends to find which each image belongs to a race class. This confidence is
a M-dimensional subspace whose basis vectors correspond helpful to the image-based face recognition, and cross-race
to the maximum variance direction in the original image space. face recognition. Separating the race factor from the other
36
© 2011 ACEEE
DOI: 01.IJIT.01.02. 214
 ACEEE Int. J. on Information Technology, Vol. 01, No. 02, Sep 2011

factors can help the recognition system to extract more ACKNOWLEDGMENT

identity-sensitive features, thereby enhancing the
The authors would like to thank Dr. Yunhong Wang for
performance of the current face identity recognition systems.
providing us the NLPR database.
The proposed system can also be applied to other image-
based classification tasks. Future experiments using racially
REFERENCES
ambiguous faces need to involve participants of other races.
[1] P.Jonathon Phillips, Fang jiang, Abhijit Narvekar, Julianne
V. EXPERIMENTAL RESULTS Ayyad, and Alice J.C’ Toole, “An Other-Race Effect for Face
Recognition Algorithms”.
A face image data set is collected from Internet, and [2] Serena Holguin, dawn E. McQuiston, Otto H. MacLin, and
divided into a training dataset and a test dataset. Using face Roy S. Malpass, “Racial Classification of Racially Ambiguous
detector and face alignment tool, these faces are automatically Faces”, Arpil 13, 2000.
cropped and normalized in grey level and geometry as in [6], [3] R. Malpass and J. Kravitz, “ Recognition for faces of own and
and each face is manually labeled with an age value estimated other rae”, pp.330-334,1969.
by human subjectively. The dataset is separated into two [4] J. Brigham and P. Barkowitz, “Do’they all look alike?’ the
effect of race, sex, experience and attitudes on the ability to recognize
race groups, Myanmar and Non-Myanmar. The Non-Myanmar
faces,” J. Appl. Soc. Phychol 8, pp, 306-318, 1978.
database is composed of research papers. Most of the [5] A Golby, J. Gabrieli, J. Chiao and J.Eberhardt, “Differential
Myanmar faces are of Kachin, Kayah, etc. origins. These responses in the fusiform region to same-race and other-race faces,”
face images are contained variations in pose, illumination Nautre Neuroscience 4(8), pp. 845-850, 2001.
and expression. Sample images from the databases are shown [6] A . Puce, T.Allison, J. Gore and G. McCarthy, “Face-sensitive
in Fig. 3. regions in human extrastriate cortex studied by functional mri,” J.
Neurophysicol. 74, pp. 1192-1199,1995.
[7] B. Moghaddam and M. Yang, “Learning gender with support
faces,” IEEE trans. Pattern Analysis and Machine Intelligence 24,
pp. 707-711, May. 2002.
[8] B.Glolomb, D. Lawrence, and T.Sejnowski, “Sexnet: A neural
network identifies sex from human faces,” in Advances in Neural
Information Processing Systems (NIPS), 3, pp. 572-577, 1990.
[9] S.Gutta, J. Huang, P. Phillips, and H. Wechsler, “Mixture of
experts for classification of gender, ethnic origin, and pose of human
faces”, IEEE Trans. Neural Networks 11, pp. 948-960, Jul.2000.
[10] G.Shakhnarovich, P.a.viola, and B. Moghaddam, “A unified
learning framework for real time face detection and classification,”
in Proc. IEEE International Conference on Automatic Face and
Gesture Recognition, 2002.
[11] R.K. Bothwell, J. C. Brigham , and R.S. Malpass, “Cross-
racial identification,” Personality & Social Psychology Bulletin, Vol.
15,pp. 19-25, 1989.
[12] C.A. Meissner and J. C .Brigham, “Thirty years of investigating
the own-race bias in memory for faces: A meta-analytic review,”
Psychology, Public Policy, and Law, vol.7, pp. 3-35, 2001.
[13] D.J.Kelly, P.C. Quinn, A. M. Slater, K.Lee, L.. Ge and O.P
ascalis, “The other-race effect develops during infancy: Evidence
of perceptual narrowing,” Psychological Science, vol. 18, pp. 1084-
Figure 3. Representative faces in the database . (a) Myanmar; (b) 1089,2007.
Non-Mya nma r. [14] M . J . Er , W . Chen , S . Wu “ High Speed Face Recognition
based on discrete cosine transform and RBF neural network” IEEE
Trans on Neural Network Vol . 16 (2007), No . 3 , PP . 679,691.
[15] Kwon, Y.H. and da Vitoria Lobo, N.1993. “Locating Facial
Features for Age Classification”, In Proceedings of SPIE-the
International Society for Optical Engineering Conference. 62-72.
[16] Hayashi, J., Yasumoto, M., Ito, H., Niwa, Y.and Koshimizu,
H.2002. Age and Gender Estimation from Facial Image Processing.
In Proceedings of the 41 st SICE Annual Conference. 13-18
[17] Lanitis, A. 2002. On the Significance of Different Facial Parts
for Automatic Age Estimation. In Proceedings of the 14 the
International Conference on Digital Signal Processing. 1027-1030

Figure 4. Performance comparison

37
© 2011 ACEEE
DOI: 01.IJIT.01.02. 214
View publication stats