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

net/publication/228790347

A Summary of literature review: Face Recognition

Article · August 2007

CITATIONS READS

9 20,901

2 authors:

Tt Tt Dzulkifli Mohamad
Suan Sunandha Rajabhat University Universiti Teknologi Malaysia
5 PUBLICATIONS   62 CITATIONS    106 PUBLICATIONS   1,264 CITATIONS   

SEE PROFILE SEE PROFILE

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

Red Blood Cell View project

Pattern Recognition View project

All content following this page was uploaded by Dzulkifli Mohamad on 30 September 2014.

The user has requested enhancement of the downloaded file.

Postgraduate Annual Research Seminar 2007 (3-4 July 2007) 1

A Summary of literature review : Face Recognition

Kittikhun Meethongjan & Dzulkifli Mohamad

Faculty of Computer Science & Information System,
University Technology of Malaysia, 81310 Skudai, Johor, Malaysia.
kittmee@yahoo.com; dzulkifli@fsksm.utm.my

Abstract recognition can be done in two formulate both a still

Approach of face recognition aims to detect
faces in still image and sequence image from video
have many method such as local, global, and hybrid
approach. The main problem of face recognition are
intensity, illumination, pose, difficult to controlling
and large occlusion. In 3D capture creates larger data
files per subject which applies significant storage
requirements, slow processing, most new devices can
be capture in 3D. This is the problem for our future
work that want to solving and create accuracy gain
for widely accept in 3D face recognition system.
Keyword : face recognition, face expression,
literature review
image and video image of a scene. There have
divided into two basic application: identification and
verification. In the identification problem, the face to
be recognized is unknown and matched against face
of a database containing known individuals. In the
verification problem the system confirms or rejects
the claimed identify of the input face.[3]-[5]
However, before face recognition is performed. the
system should determine whether or not there is a
face in a given image or given video, a sequence of
images. This process is called face detection. Once a
face is detected, face region should be isolated from
the scene for the face recognition. The overall
process is depicted in Fig 1[6],[39]

1. Introduction
Face recognition is one of the few biometric
methods that possess the merits of both high accuracy
and low intrusiveness. It has the accuracy of a
physiological approach without being intrusive. Over
past 30 years, many researchers have been proposed
different face recognition techniques, motivated by
the increased number of real world applications
requiring the recognition of human faces. There are
several problems that make automatic face
recognition a very difficult task. However, the face
image of a person inputs to the database that is
usually acquired under different conditions. The Figure 1. Configuration of a generic face
important of automatic face recognition is much be recognition
cope with numerous variations of images of the same
face due to changes in the following parameters such
as pose, illumination, expression, motion, facial hair, 2. Face recognition approach
glasses, and background [1][2]
Face recognition technology is well advance that Face recognition can be done in both a still
can applied for many commercial applications such image and video sequence which has its origin in
as personal identification, security system, image- still-image face recognition. Different approaches of
film processing, psychology, computer interaction, face recognition for still images can be categorized
entertainment system, smart card, law enforcement, into tree main groups such as holistic approach,
surveillance and so on. A general problem of face feature-based approach, and hybrid approach [7]:

This paper has not been revised and corrected according to reviewers comments Copyright PARS’07.
Postgraduate Annual Research Seminar 2007 (3-4 July 2007) 2

2.1 Holistic Approach representation in figure 2(b), each derived from a

different face image.
In holistic approach or global feature, the whole
face region is taken into account as input data into
face detection system. Examples of holistic methods
are eigenfaces (most widely used method for face
recognition), probabilistic eigenfaces, fisherfaces,
support vector machines, nearest feature lines (NFL)
and independent-component analysis approaches.
They are all based on principal component-analysis
(PCA) techniques that can be used to simplify a
dataset into lower dimension while retaining the
characteristics of dataset.

2.2 Feature-based Approach (a) Elastic graph representation

In feature-based approaches or local feature that is

the features on face such as nose, and then eyes are
segmented and then used as input data for structural
classifier. Pure geometry, dynamic link architecture,
and hidden Markov model methods belong to this
category. One of the most successful of these systems
is the Elastic Bunch Graph Matching (EBGM)
system [40],[41], which is based on DLA. Wavelets,
especially Gabor wavelets, play a building block role
for facial representation in these graph matching
methods. A typical local feature representation
consists of wavelet coefficients for different scales
and rotations based on fixed wavelet bases. These
locally estimated wavelet coefficients are robust to (b) Bunch graph
illumination change, translation, distortion, rotation,
and scaling. [42], [43]. The grid is appropriately Figure 2. The bunch graph representation of
positioned over the image and is stored with each faces used in elastic graph matching[41]
grid point’s locally determined jet in figure 2(a), and
serves to represent the pattern classes. Recognition of
a new image takes place by transforming the image 2.3 Hybrid Approach
into the grid of jets, and matching all stored model
graphs to the image. Conformation of the DLA is The idea of this method comes from how human
done by establishing and dynamically modifying vision system perceives both holistic and local
links between vertices in the feature. The key factors that influence the
model domain. performance of hybrid approach include how to
DLAs attempt to solve some of the conceptual determine which features should be combined and
problems of conventional artificial neural networks, how to combine, so as to preserve their advantages
the most prominent of these being the representation and avert their disadvantages at the same time. These
of syntactical relationships in neural networks. DLAs problems have close relationship with the multiple
use synaptic plasticity and are able to form sets of classifier system (MCS) [29] and ensemble learning
neurons grouped into structured graphs while [30] in the field of machine learning. Unfortunately,
maintaining the advantages of neural systems. The even in these fields, these problems remain unsolved.
DLA architecture was recently extended to Elastic In spite of this, numerous efforts made in these fields
Bunch Graph Matching in figure 10[41] This is indeed provide us some insights into solving these
similar to the graph described above, but instead of problems, and these lessons can be used as guidelines
attaching only a single jet to each node, the authors in designing a hybrid face recognition system. For
attached a set of jets that can show bunch graph example, components of a hybrid system, either

This paper has not been revised and corrected according to reviewers comments Copyright PARS’07.
Postgraduate Annual Research Seminar 2007 (3-4 July 2007) 3

feature or classifier, should be both accurate and 3. Review Model Method

diverse, such that a complementary advantages can
be feasible. 3.1 Hidden Markov Models Method
In fact, local features and global features have
quite different properties and can hopefully offer Hidden Markov models (HMM) is another
complementary information about the classification promising method that works well for images with
task. Table 1 summarizes qualitatively the difference variation in different lighting, facial expression, and
between the two types of features. We can see from orientation. HMM is a set of statistical models used
the table that local features and global ones are to characterize properties of signals. It has very good
separately sensitive to different variation factors. For performance in speech recognition and character
instance, illumination changes may have more recognition, where the data is 1-dimentional. The
influence on local features, while expression changes system being modeled is assumed to be a Markov
have more impact on holistic features. For these process with unknown parameters, and the goal is to
observations, hybrid approach that use both holistic find hidden parameters from the observable
and local information for recognition may be an parameters. The each state in HMM has a probability
effective way to reduce the complexity of classifiers distribution over the possible output whereas each
and improve their generalization capability. state in a regular Markov model is observable. In
Nefian’s paper [8], the authors use HMM approach
Table 1 Comparison of the local features and global for face recognition based on the extraction of 2-
features’ sensitiveness to variations dimensional discrete cosine transformation (DCT)
feature vectors. The author takes advantage of DCT
Variation Local Holistic compression property for feature extraction. An
factors features features image is divided by blocks of a sub-image associated
Small Var. not sensitive sensitive with observation vector. More details about HMM
method are provided in the following sections.
Large Var. sensitive very sensitive In HHM, there are unobservable Markov chain
with limited number of status in the model, the
Illuminations very sensitive sensitive observation symbol probability matrix B, a state
[31] transition probability matrix A, initial state
Expressions not sensitive sensitive distribution π, and set of probability density functions
[32],[33] (PDF). A HMM is defined as the triplets λ = (A, B,
Pose [34] sensitive very sensitive π).
Noise [35] very sensitive Sensitive For frontal human face images, the important facial
components appear in top to bottom order such as
Occlusion not sensitive very sensitive hair, forehead, eyes, nose, mouth, and chin. This still
[32],[33] holds although the image rotates slightly in the image
plane. Each of the facial region is assigned to one
state in 1-D continuous HMM. The transition
Despite the potential advantages, the work in this probability ai j and structure of face model is
the category is still relatively few, possibly due to the illustrated in Fig. 3 [9].
difficulties mentioned above, while typical hybrid
approach in traditional sense such as flexible
appearance models [36], hybrid LFA [37], are
generally not suitable for handling the one sample
problem. We hope more researching efforts could be
engaged in this approach, and in doing so, we believe
that the potential power of hybrid approach would put
forth sooner or later. [38]

In this paper, we review some method from three Figure 3 : HMM for face recognition
approach such as Hidden Markov Models Method,
Neural Network Method, Eigen Face Method, and Each face image with W and height H is divided
Fisher face Method. into overlapping blocks of height L and the same

This paper has not been revised and corrected according to reviewers comments Copyright PARS’07.
Postgraduate Annual Research Seminar 2007 (3-4 July 2007)
width. The block extracting is shown in Fig. 4 [9].
The amount of the overlapping P has a significant
effect on recognition rate since features are captured
independent of vertical position. The magnitude of L
is also important. Small length of L will assign
insufficient information to discriminate to the
observation vector. On the other hand, large value of
L will increase the chances of cutting across the
feature. Therefore, it is important to find good value
for L. Once blocks are extracted from the image, a set
of DCT coefficients are calculated for each block.
When each block is transformed with DCT, the most
important coefficients with low frequencies are
converged and clustered in small area in the DCT
domain. The author from [9] uses 12x3 size window
to pick these significant information of signal energy.
In this way, the size of observation vector is reduced
significantly, which makes the system very efficient
while still retaining good detecting rate. In the
training phase, the image is segmented from top to
bottom where each segment corresponds to a state,
and initial observation probability matrix B is
obtained from observation vectors associated with
each state. Once B is obtained, the initial value of A
and π are set given the left to right structure of the
face
Figure 4. Block extraction from image
The face image is recognized if given Markov
mode, the probability of observation symbols is
maximum. For experiment in the paper, 400 images
of 40 individuals with 10 face images per individual
are used. The image database contains face images
with different expressions, hair styles, eye wears and
head orientations. The system achieves 84 % correct
classification with L =10 and P =9 while eigenfaces
approach achieves 73% of correct classification with
the same dataset. Considering this fact, HMM
approach has a bit better performance than eigenfaces
method for images with variations.
This paper has not been revised and corrected according to reviewers comments Copyright PARS’07.
4
3.2 Neural Network Method
Neural networks-based approaches are learned
from the example-images and rely on the techniques
from machine learning to find the relevant
characteristics of face images. The learned
characteristics, in the form of discriminant functions
(i.e. non-linear decision surfaces), are subsequently
used for face recognition. Conventionally, face
images are projected to a low-dimensional feature
space and nonlinear decision surface is formed using
multilayer neural networks for classifications and
recognition [10]. Neural networks have also been
used successfully for face recognition problem
[11],[12],[10]. The advantage of using the neural
networks for face recognition is that the networks can
be trained to capture more knowledge about the
variation of face patterns, and thereby achieving good
generalization [13]. The main drawback of this
technique is that the networks have to be extensively
tuned to get exceptional performance. Among the
neural networks approaches for face recognition,
multilayer perceptron (MLP) with back propagation
(BP) algorithm has been mostly used [14]. However,
the convergence of the MLP networks is slow and the
global minima of the error space may not be always
achieved [11]. On the other hand, the RBF neural
networks have fast learning ability [15] and best
approximation property [16]. So, in recent times,
many researches have used RBF networks for face
recognition and show in figure 5 [17],[18],[13].
However, their success rates are not so
promising as the error rates vary from 5 to 9% under
variation of pose, orientation, scale and light [13].
This may be due to the fact that the selection of the
centers of the hidden layer neurons might not have
been done by capturing the knowledge about the
distribution of training patterns and variations of face
pose, orientation and lighting. [19]
Figure 5. Structure of RBF neural network
Postgraduate Annual Research Seminar 2007 (3-4 July 2007) 5

3.3 Eigenface Method distance for example). The class of the most similar
vector is the result of the recognition process, i.e. the
In 1991, Turk and Pentland used PCA identity of the face. In addition, a Rejection System
projections as the feature vectors to solve the problem for unknown faces is used if the similarity matching
of face recognition, using the Euclidean distance as measure is not good enough and show in figure 6 [24]
similarity function [20]. This system, later called
Eigenfaces, was the first eigenspace-based face
recognition approach and, from then on, many
eigenspace-based systems have been proposed using
different projection methods and similarity functions.
In particular, Belhumeur et al. proposed in 1997 the
use of FLD as projection algorithm in the so-called
Fisherfaces system [21]. In all standard eigenspace-
based approaches a similarity function, which works
as a nearest-neighbor classifier [22], is employed.
In 1997, Pentland and Moghaddam proposed a Figure 6. Block diagram of a generic eigenspace-
differential eigenspace-based approach that allows based face recognition system
the application of statistical analysis in the
recognition process [23]. The main idea is to work
with differences between face images, rather than 3.4 Fisherface Method
with single face images. In this way the recognition
problem becomes a two-class problem, because the Fisherface algorithm considers the ratio between
so-called “differential image” contains information of the variation of one person and that of another
whether the two subtracted images belong to the person. It maximizes the determinant of between-
same class or to different classes. In this case the class scatter matrix simultaneously, minimizing the
number of training images per class increases so that determinant of within-class scatter matrix.
statistical information becomes available, and a For a C – class problem, the between -class
statistical classifier can be used for performing the scatter matrix is defined as follows:
recognition. The system proposed in [23] used Dual-
PCA projections and a Bayesian classifier.
Eigenspace-based approaches approximate the
face vectors (face images) by lower dimensional
feature vectors. These approaches consider an off-
line phase or training, where the projection matrix where Pr(Ωi ) is the prior class probability, µi is the
(W ∈ RN ×m), the one that achieve the dimensional mean sample of class Ωi and µ is the mean sample
reduction, is obtained using all the database face of all classes.
images. In the off-line phase, the mean face ( x ) and The within-class scatter matrix is defined as
the reduced representation of each database image follows:
( pk ) are also calculated. The recognition process
works as follows. A preprocessing module transforms
the face image into a unitary vector (normalization
module in the case of Fig. 1) and then performs a
subtraction of the mean face. The resulting vector is
where
projected using the projection matrix that depends on
the eigenspace method been used (PCA, FLD, etc.).
This projection corresponds to a dimensional is covariance matrix of within-class sample. N i is
reduction of the input, starting with vectors in RN the number of samples in class Ωi .
(where N is the dimension of the image vectors) and
obtaining projected vectors q in Rm , with m<N Fisher criteria function is defined as follows:
(usually m<<N). Then, the similarity of q with each
of the reduced vectors € pk ( € pk ∈ Rm ) is computed
using a certain criterion of similarity (Euclidean

This paper has not been revised and corrected according to reviewers comments Copyright PARS’07.
Postgraduate Annual Research Seminar 2007 (3-4 July 2007)
Then the projective matrix W fld can be chosen
as follows:
W fld can be calculated by solving the generalized
eigenvalue problem:
In face recognition applications, because rank of
S w ∈ R m×n is at most N - c, where N is the number
of images in training set and typically much smaller
than , number of pixels in each image, the within-
class scatter matrix S w is always singular. To
overcome this problem, PCA is first utilized to
reduce the dimension of the images from N to N-c,
then recalculated S w will be non-singular and FLD
can be utilized to find the projective matrix W fld ,
which is referred to as Fisherfaces and Figure 7 is a
comparison of PCA and FLD for a two-class problem
in which the samples from each class are randomly
perturbed in a direction perpendicular to a linear
subspace.[25] - [28]
Figure 7. A comparison of principal component
analysis (PCA) and Fisher’s linear
discriminant (FLD) [28]
This paper has not been revised and corrected according to reviewers comments Copyright PARS’07.
6
4. Face Expression Recognition
Several research about facial expression
recognition. The facial expressions under
examination were defined by psychologists as a set of
six basic facial expressions (anger, disgust, fear,
happiness, sadness, and surprise) [44]. In order to
make the recognition procedure more standardized, a
set of muscle movements known as Facial Action
Units (FAUs) that produce each facial expression,
was created, thus forming the so-called Facial Action
Coding System (FACS) [45]. These FAUs are
combined in order to create the rules responsible for
the formation of facial expressions as proposed in
[46].
A survey on the research made regarding facial
expression recognition can be found in [47] and [48].
The approaches reported regarding facial expression
recognition can be distinguished in two main
directions, the feature-based ones and the template-
based ones, according to the method they use for
facial information extraction. The feature-based
methods use texture or geometrical information as
features for expression information extraction. The
template-based methods use 3-D or 2-D head and
facial models as templates for expression information
extraction. However, in overview about 2D and 3D
face recognition algorithm that can be summary in
table 2:
Tanaka et al. [49] also perform curvature-based
segmentation and represent the face using an
Extended Gaussian Image (EGI). Recognition is then
performed using a spherical correlation of the EGIs.
Hesher et al. [50] explore PCA type approaches using
different numbers of eigenvectors and image sizes.
The image data set used has 6 different facial
expressions for each of 37 subjects. The performance
figures reported result from using multiple images
per subject in the gallery. This effectively gives the
probe image more chances to make a correct match,
and is known to raise the recognition rate.
Medioni et al. [51] perform 3D face recognition using
iterative closest point (ICP) matching of probe face
surface against gallery face surface. Whereas most of
the works covered here acquired 3D using structured light,
this work uses a stereo-based system. An Equal Error
Rate (EER) of ”better than 2%” is reported.
Postgraduate Annual Research Seminar 2007 (3-4 July 2007) 7

Reference number of number of images image size Reported Handles handles

persons performance size performance
variation
Approaches Using 3D Without 2D
Tanaka [49]1998 37 37 256 x 256 100 % N N
Hesher [50]2003 37 222(6 expr. ea) 242 x 347 97 % Y N
Medioni [51]2003 100 700(7 poses ea.) ? 98 % Y N
Moreno [52]2003 60 420(3 expr., 2 pose) 2186 avg. 78 % Y Some
Lee [53]2003 35 70 320 x 320 94 % rank 5 Y N
Bai [54]2005 40 on ORL 800 ? 76.3 % rank 1 N Y
20 on UMIST 575 ? 85 %
Passalis [55]2007 466 4007 640 x 480 92.5 % Y N
multi-modal approaches using a combination of 3D and 2D
Lao [56]2000 10 360 486 x 60 91 % Y N
Beumier [57]2000 29 29 gallery, 26 probes ?, 240 2D 98 % Y N
(2% EER)
Wang [58]2002 50 300 128 x 512 > 90 % N Y
Bronstein [59]2003 157 ? 2250 avg. Not report Y Y
Chang [60]2003 200, 951 480 x 640 99 % Y N
(+ 75 ingtraining)
Mian [61] 2007 466 4007 480 x 640 99.02 Y N

Table 3. Summary Of Research On 3D and Multi-Modal 2D+3D Face Recognition

Moreno and co-workers [52] approach 3D face
recognition by first performing a segmentation based
on Gaussian curvature and then creating a feature vector
based on the segmented regions. The results on a
dataset of 420 face meshes representing 60 different
persons, with some sampling of different expressions
and poses for each person. They report 78% rank-one
recognition on the subset of frontal views, and 93%
overall rank-five recognition.
Lee and co-workers perform 3D face recognition by
locating the nose tip, and then forming a feature
vector based on contours along the face at a sequence of
depth values [53]. They report 94% correct recognition
at rank 5, and do not report rank-one recognition. Given
the relatively small dataset, 35 persons, and the recognition
rates reported for other works, it would appear that the
contour-oriented method is not as powerful as other
methods.
Bai et al. [54] used extended fisherface with 3D
morpheble model is utilized to derive multiple
images form a single example image to form the
training set for Fisherface. Experiments on ORL and
UMIST face database demonstrate impressive
performance improvement of our method over the
conventional benchmarks, for face recognition
training from one image while testing under some
expressions, illumination and slight pose variations.
Passalis et al. [55] perform face recognition by
intraclass retrieval of nonrigid 3D objects. A novel
3D object retrieval method is presented which uses a
parameterized annotated model of the shape of the
class objects, incorporating its main characteristics
and transformed into the wavelet domain. The result
of this method does not require user interaction,
achieves high accuracy, efficient for use with large
databases, and suitable for nonrigid object classes.
They report 95.2% at 10_3 false accept rate on
Grand Challenge v2 database, yielding an average
verification.
Lao et al. [56] perform 3D face recognition using
a sparse depth map constructed from stereo images.
Iso-luminance contours are used for the stereo
matching. Both 2D edges and iso-luminance contours
are used in finding the irises. In this limited sense
with multi-modal approach. The report 87% to 96%

This paper has not been revised and corrected according to reviewers comments Copyright PARS’07.
Postgraduate Annual Research Seminar 2007 (3-4 July 2007)
are using a dataset of ten persons, with four images taken
at each of nine poses for each person. However, no
attempt is made to deal with variation in facial
expression.
Beumier and Acheroy [57] perform multi-modal
recognition by using a weighted sum of the 3D and
2D similarity measures. They report on experiments
with a dataset of 26 persons in the gallery and 29 persons
in the probe set, achieving recognition performance as high
as 2% equal error rate (EER) for multi-modal
recognition, compared to 4% for 3D alone and 8% for
2D alone. They found that 3D+2D data was acquired
on a larger set of 120 persons on each of two
different acquisition sessions.
Wang et al. [58] use Gabor filter responses in 2D and
“point signatures” in 3D to perform multi-modal face
recognition. The 2D and 3D features together form a
feature vector and used support vector
machines(SVM). Experiments were performed with
images from 50 subjects, six images per subject, with
pose and expression variations. Recognition rates
exceeding 90%.
Bronstein et al. [59] used an isometric
transformation approach to 3D face analysis in an
attempt to better cope with variation in face shape
due to facial expression. One method they propose is
effectively multi-modal 2D+3D recognition using
Eigen composition of fattened textures and canonical
images. They show examples of correct and incorrect
recognition by different algorithms, but do not report
any overall quantitative performance results for any
algorithm.
Chang et al. [60] report on PCA-based recognition
experiments performed using 3D and 2D images from
200 persons. One experiment uses a single set of later
images for each person as the probes, and another
experiment uses a larger set of 676 probes. Results in
both experiments were approximately 99% rank-one
recognition for multi-modal 3D+2D, 94% for 3D
alone and 89% for 2D alone. The combined result
was obtained using a weighted sum of the distances
from the individual 3D and 2D face spaces. This
work represents the largest experimental study yet
reported in the literature either for 3D face alone or
for multi-modal 2D+3D.
This paper has not been revised and corrected according to reviewers comments Copyright PARS’07.
8
Mian et al. [61] used multimodal (2D and 3D)
and performs hybrid (feature-based and holistic)
matching to achieve efficiency and robustness to
facial expressions. A novel 3D Spherical Face
Representation (SFR) is used in conjunction with the
SIFT descriptor which quickly eliminates a large
number of candidate faces at an early stage for
efficient recognition in case of large galleries. This
approach automatically segments the eyes-forehead
and the nose regions. The results of all the matching
engines are fused at the metric level to achieve higher
accuracy. However, multimodal hybrid algorithm
better than others approach and identification rates of
99.02% and 95.37% for probes with neutral and non-
neutral expression respectively.
5. Summary of review
The approach for 3D face recognition involves
sensitivity to size variation that can be use a purely
curvature-based representation and handle size
change between faces, but run into problems with
change of facial expression between the enrollment
image and the image to be recognized. In facial
recognition system should be able to handle variation in
expression. The seriousness of the problem of
variation in facial expression between the enrollment
image and the image to be recognized is illustrated in
the results shown in Figure 2. This experiment
focuses on the effects of expression change.
Recognition was done with PCA-based 2D and 3D
algorithm. The upper cumulative match characteristic
(CMC) curves represent performance with time lapse
only between gallery and probe.
A main problem to experimental validation and
comparison of 3D face recognition is lack of
appropriate datasets. Desirable properties of such a
dataset include: (1) a large number and demographic
variety of people represented, (2) images of a given
person taken at repeated intervals of time, (3) images
of a given person that represent substantial variation
in facial expression, (4) high spatial resolution, for
example, depth resolution of 0.1 mm or better, and (5) low
frequency of sensor-specific artifacts in the data.
Postgraduate Annual Research Seminar 2007 (3-4 July 2007)
6. Future work
In reviewing last research, we found many
approach can be used for face recognition that each
method have different advantage and disadvantage
such as local, global, and hybrid method. There are
have 2 type of image for face recognition technique:
still image and video image (still image sequence).
However, we found some problems in face
recognition system such as: (1) pose problem : due to
can not control face image for capturing and have
many pose variation will be change every time. (2)
illumination problem : due to source image have
light condition or different lighting and viewing
variations. (3) environment problem : due to in fact,
motion and expression can not controlling that is
natural image. (4) 3D problem : due to in 3D image
must be used more storage, many variable, low
speed, and lack of face database testing.
In future work, we propose a novel method for
face recognition by hybrid approach combines 3D
face and face expression. (eyes, nose, and mouth are
location feature for extraction). Due to our survey
found 3D recognition more accurate than 2D
recognition, 3D capture creates larger data files per
subject which applies significant storage
requirements, slow processing, most new devices can
be capture in 3D, and can not control environment
from the real world. Lastly, we want to widely
accepted in 3D such as 3D Morpheble Model
approach can be recognize both frontal face and non-
frontal face image. This is idea to use for improving
overall recognition performance.
7. References
[1] Lu J., Yuan X., and Yahagi T., A method of Face
recognition based on Fuzzy c-Means clutering and
associated sub-NNs. Proc. IEEE, 18(1), 2007, 150-159.
[2] Kelly M. D., Visual Identification of People by
Computer. Stanford AI Project, Stanford, CA, Technical
Report, 1970, AI-130.
This paper has not been revised and corrected according to reviewers comments Copyright PARS’07.
9
[3] Chellappa R., Wilson C.L., and Sirohey S., Human
and Machinr Recognition of Faces: A Survey, Proc. IEEE,
(83), 1995, 705-741.
[4] Zhang J., Yan Y., and Lades M., Face Recognition:
Eigenfaces, Elastic Matching, and Neural Nets, Proc.
IEEE, l.85(9), 1997, 1423-1435.
[5] Torres L., Is there any hope for face recognition.
Technical University of Catalonia, Spain, 2004, 1-4.
[6] Kim H. H., Survey paper: Face Detection and Face
Recognition. Department of Computer science, University
of Saskatchewan, South Korea, 2004, 1-7.
[7] Zhao W., Chellappa R., Phillips P. J., and
Rosenfeld A., Face recognition: A literature survey,
ACM Computing Surverys (CSUR), 35(4), 2003, 399-
458.
[8] Nefian A.V., and al et., Hidden markov models
for face recognition, In Proceedings, International
Conference on Acoustics, Speech and Signal
Proceeding, 1998, 2721-2724.
[9] Turk M. A., and Pentland A. P., Eigenfaces for
recognition, Cognitive Neurosci, 3(1), 1991, 71-86.
[10] Rowley H., Baluja S., and Kanade T., Neural
network-based face detection, IEEE Trans. Pattern
Anal. Mach. Intell, 20, 1998, 23-38.
[11] Er M.J., Wu S., Lu J., Toh H.L., Face
recognition with radial basis function (RBF) neural
networks, IEEE Trans. Neural Net,13, 2002, 697-
710.
[12] Osuna E., Freund R., Girosi F., Training support
vector machines: an application to face detection, in:
Proceedings of the IEEE Conference Computer
Vision and Pattern Recognition,1997, 130-136.
[13] Yang F., Paindovoine M., Implementation of an
RBF neural network on embedded systems: real-time
face tracking and identity verification, IEEE Trans.
Neural Network, (14), 2003, 1162-1175.
[14] Valentin D., Abdi H., O’Toole A.J., and Cottrell
G.W., Connectionist models for face processing: a
survey, Pattern Recognition, 27, (1994), 1209–1230.
Postgraduate Annual Research Seminar 2007 (3-4 July 2007)
[15] Moody J., and Darken C.J., Fast learning in
network of locallytuned processing units, Neural
Computing, vol 1, page 281–294, 1989.
[16] Girosi F., and Poggio T., Networks and the best
approximation property, Biol. Cybern, (63), 1990,
169-176.
[17] Howell, H. Buxton, Learning identity with
radial basis function networks, Neurocomputing, l 20,
1998, 15-34.
[18] Ranganath S., and Arun K., Face recognition
using transform features and neural networks, Pattern
Recognition, l 30, 1997, 1615-1622.
[19] Sing J. K., Basu D. K., Nasipuri M., and Kandu
M., Face recognition using point symmetry distance-
based RBF network, Applied soft computing,
available at www.sciencedirect.com, l 7, January
2007, 58-70.
[20] Turk M. and Pentland A., Eigenfaces for
Recognition, J. Cognitive Neuroscience, 3(1), 1991,
71-86.
[21] Belhumeur P.N., Hespanha J.P., and Kriegman
D.J., Eigenfaces vs. Fisherfaces: recognition using
class specific linear projection, IEEE Trans. Pattern
Analysis and Machine Intelligence, 19(7), July 1997,
711-720.
[22] Duda R.O., Hart P.E., and Stork D.G., Pattern
Classification, Second Edition, 2001.
[23] Pentland A. and Moghaddam B., Probabilistic
Visual Learning for Object Representation, IEEE
Trans. Pattern Analysis and Machine Intelligence,
19(7), July 1997, 696-710.
[24] Ruiz J. D. S., and Navarrete P., Eigenspace-
based Face Recognition: A comparative study of
different approaches, IEEE Trans. on Sys., Man. &
Cyb. C., 16(7), 2002, 817-830.
[25] Belhumeur P. N., Hespanha J. P., and Kriegman
D. J., Eigenfaces vs. Fisherfaces: Recognition using
class specific linear projection, IEEE Trans. Pattern
Anal. Machine Intel, 19, May 1997, 711-720.
[26] Bai X. M., Yin B. C., Shi Q., and Sun Y. F.,
Face recognition using extended Fisherface with3D
This paper has not been revised and corrected according to reviewers comments Copyright PARS’07.
10
Morphable model, Proceedings of the Fourth
International Conference on Machine Learning and
Cybernetics, Guangzhou, August 2005, 18-21.
[27] Duda R., and Hart P., Pattern Classification and
Scene Analysis. New York: Wiley, 1973.
[28] Belhumeur P. N., Hespanha J. P., and Kriegman
D. J., Eigenfaces vs. Fisherfaces: Recognition Using
Class Specific Linear Projection, IEEE Transactions
on pattern analysis and machine intelligence, 19(7),
July 1997.
[29] Kittler J., Hatef M., Duin R.P.W., and Matas J.,
On combining classifiers. IEEE Trans. Pattern
Analysis and Machine Intelligence, 20(3), 1998, 226-
239.
[30] Zhou Z.H., Wu J., and Tang W., Ensembling
neural networks: Many could be better than all.
Artificial Intelligence, 137(1-2), 2000, 239-263.
[31] Hallinan P.W., and al. et, Two-and three-
dimensional patterns of the face, Natick, MA: A K
Peters,Letd, 1999.
[32] Martinez A.M., Recognizing imprecisely
localized, partially occluded, and expression variant
faces from a single sample per class. IEEE Trans.
Pattern Analysis and Machine Intelligence, 25(6),
2002, 748-763.
[33] Tan X., Chen S.C., Zhou Z.-H., and Zhang F.,
Recognizing partially occluded, expression variant
faces from single training image per person with
SOM and soft kNN ensemble. IEEE Transactions on
Neural. Networks, 16(2005),4875-886.
[34] Heisele, B. T., Serre, M., Pontil and Poggio, T.,
Component-based face detection. In Proceedings,
IEEE Conference on Computer Vision and Pattern
Recognition, 1(2001) 657-662.
[35] Costen N.P., Cootes T.F., Taylor C.J.,
Compensating for ensemble-specific effects when
building facial models, Image and Vision Computing,
20(2002) 673-682.
[36] Lanitis A., Taylor C.J., and Cootes T.F.,
Automatic face identification system using flexible
appear ance models. Image Vis. Comput. 13(1995)
393-401.
Postgraduate Annual Research Seminar 2007 (3-4 July 2007)
[37] Penev, P. and Atick, J. Local feature analysis: A
general statistical theory for objecct representation.
Netw.: Computat. Neural Syst. 7(1996) 477-500.
[38] Tan X., Chen S., Zhou Z. H., and Zhang F., Face
recognition from a single image per person: a survey,
Institution of Automation, Chinese Academy of
Sciences Nanjing University of Aeronautics &
Astronautics, Nanjing China. (2002) 1-34
[39] Zhao W., Chellappa R., and Rosenfeld A., Face
recognition: a literature survey. ACM Computing Surveys,
35:399-458, December 2003.
[40] Okada K., Steffans J., Maurer T., Hong H.,
Elagin E., Neven H., Andmalsburg C. V. D., (1998).
The Bochum/USC Face Recognition System and how
it fared in the FERET Phase III Test. In Face
Recognition: From Theory to Applications, Eds.
Springer-Verlag, Berlin, Germany, (1998) 186-205.
[41] Wiskott L., Fellous J.-M., and Von dermalsburg
C., (1997), Face recognition by elastic bunch graph
matching. IEEE Trans. Patt. Anal. Mach. Intell. 19,
(1997) 775-779.
[42] Buhman J., Lades M., and Malsburg C. V. D.,
(1990), Size and distortion invariant object
recognition by hierarchical graph matching. In
Proceedings,International Joint Conference on
Neural Networks. (1990) 411-416.
[43] Lades M., Vorbruggen J., Buhmann J., Lange J.,
malsburg C. V.D., Wurtz R., and Konen W., (1993),
Distortion invariant object recognition in the dynamic
link architecture. IEEE Trans.Comput. 42, (1993)
300-311.
[44] Ekman P., and W. V. Friesen, Emotion in the
Human Face. Englewood Cliffs, NJ: Prentice-Hall,
(1975).
[45] Kanade T., Cohn J., and Tian Y.,(2000),
“Comprehensive database for facial expression
analysis,” in Proc. IEEE Int. Conf. Face and Gesture
Recognition, Mar. 2000, 46-53.
[46] Pantic M., and Rothkrantz L. J. M., “Expert
system for automatic analysis of facial expressions,”
Image Vis. Comput., 18(11), Aug. 2000, 881-905
This paper has not been revised and corrected according to reviewers comments Copyright PARS’07.
11
[47] ——, “Automatic analysis of facial expressions:
The state of the art,” IEEE Trans. Pattern Anal.
Mach. Intell., 22(12), Dec. 2000, 1424 -1445
[48] Fasel B., and Luettin J., “Automatic facial
expression analysis: A survey,” Pattern Recognit.,
36(1), 2003, 259-275.
[49] Tanaka H. T., Ikeda M., and Chiaki H.,
Curvature-based face surface recognition using
spherical, Third Interna- tional Conference on
Automated Face and Gesture Recogni-tion,
1998,372-377.
[50] Hesher C., Srivastava A., and Erlebacher. A G.,
Novel technique for face recognition using range
images. Seventh Int’l Symp. on Signal Processing and
Its Applications, 2003.
[51] Medioni G. and Waupotitsch R., Face
recognition and modeling in 3D. IEEE Int’l
Workshop on Analysis and Modeling of Faces and
Gestures (AMFG 2003), October 2003, 232-233.
[52] Moreno A. B., Sanchez A., Eelez J. F.V. , and
Diaz F. J., Face recognition using 3D surface-
extracted descriptors. Irish Machine Vision and
Image Processing Conference (IMVIP 2003),
September 2003.
[53] Lee Y., Park K., Shim J., and Yi T., 3D face
recognition using statistical multiple features for the
local depth information. 16th International
Conference on Vision Interface, available at
www.visioninterface.org/vi2003, June 2003.
[54] Bai X. M., Yin B. C., Shi Q., and Sun Y. F., Face
recognition using extended fisherface with 3D morphable
model, Proceeding of the Fourth International Conference
on Machine Learning and Cybernetics, Guangzhou, August
2005, 18-21.
[55] Passalis G., Kakadiaris I. A., and Theoharis T.,
Intraclass Retrieval of Nonrigid 3D Objects:
Application to Face Recognition., IEEE Transactions
on pattern analysis and machine intelligence, 29(2),
February 2007, 218-229.
[56] Lao S., Sumi Y., Kawade M., and Tomita F.,
3D template matching for pose invariant face
recognition using 3D Facial Model Built with
 Postgraduate Annual Research Seminar 2007 (3-4 July 2007) 12

Isoluminance Line Based Stereo Vision, Int’lConf. on
Patt. Rec. (ICPR 2000), 2000, 91-916.
[57] Beumier C. and Acheroy M., Face verification
from 3D and grey level cues. Patt. Rec. Letters, 22,
2001, 1321-1329.
[58] Wang Y., Chua C., and Ho Y., Facial feature
detection and face recognition from 2D and 3D
images. Pattern Recognition Letters, 23, 2002, 1191-
1202.
[60] Chang K., Bowyer K., and Flynn P., Face
recognition using 2D and 3D facial data. 2003
Multimodal User Authentication Workshop,
December 2003, 25-32.
[61] Mian A., Bennamoun M., and Owens R., An
efficient Multimodal 2D-3D Hybrid approach to
automatic face recognition, IEEE Transactions on
pattern analysis and machine intelligence, January 10,
2007, 1-34.

[59] Bronstein A. M., Bronstein M. M., and Kimmel

R., Expression-invariant 3D face recognition. Audio-
and Video-Based Person Authentication (AVBPA
2003), LCNS 2688, 2003, 62-70.

This paper has not been revised and corrected according to reviewers comments Copyright PARS’07.

View publication stats