(IJACSA) International Journal of Advanced Computer Science and Applications,

Vol. 9, No. 6, 2018

Study of Face Recognition Techniques: A Survey

Madan Lal, Kamlesh Kumar Rafaqat Hussain Arain, Abdullah Maitlo,
Department of Computer Science Sadaquat Ali Ruk, Hidayatullah Shaikh
Sindh Madressatul Islam University, Department of Computer Science, Shah Abdul University,
Karachi, Sindh, Pakistan Khairpur, Sindh, Pakistan

Abstract—With the rapid growth in multimedia contents, same person in the scene, and face identification who is this
among such content face recognition has got much attention person in that scene. In the first phase it locates a face in an
especially in past few years. Face as an object consists of distinct image. Similarly, in the second stage, it extracts features from
features for detection; therefore, it remains most challenging an image for discrimination. After that they are matched with
research area for scholars in the field of computer vision and face database images in order to recognize correct face image
image processing. In this survey paper, we have tried to address as shown in Fig. 1. However, some existing recognition
most endeavoring face features such as pose invariance, aging, methods for authentication suffer from lack of reliability. For
illuminations and partial occlusion. They are considered to be instance, smart cards, wallets, keys, tokens use PINs and
indispensable factors in face recognition system when realized
passwords which are very difficult to remember. In addition,
over facial images. This paper also studies state of the art face
detection techniques, approaches, viz. Eigen face, Artificial
these passwords and codes can easily be forgotten; also these
Neural Networks (ANN), Support Vector Machines (SVM), magnetic cards can be misplaced, even robbed or reproduced.
Principal Component Analysis (PCA), Independent Component As a result this makes them illegible. Contrary to biological
Analysis (ICA), Gabor Wavelets, Elastic Bunch Graph Matching, characteristics and traits of an individual person they cannot
3D morphable Model and Hidden Markov Models. In addition to be stolen, forgotten or misplaced [1].
the aforementioned works, we have mentioned different testing Biometric recognition system can be built through various
face databases which include AT & T (ORL), AR, FERET, LFW,
techniques. However, most commonly used are fingertips and
YTF, and Yale, respectively for results analysis. However, aim of
iris methods. These require individual’s participation or
this research is to provide comprehensive literature review over
face recognition along with its applications. And after in depth
involvement to access the system. Moreover, latest systems
discussion, some of the major findings are given in conclusion. provide participant access without its intervention. Among
such methods face recognition is one of the most viable
Keywords—Face recognition; illuminations; partial occlusion; technique in which an individual can easily be captured and
pose invariance monitored through the system. Face recognition databases
vary with controllable photos to uncontrollable videos, where
I. INTRODUCTION LFT is used for controllable photos and YTM is used for
st
The 21 century is a modern and scientific era in which a videos. Face recognition system comprises of three main
lot of progress has been achieved as to expedite humans for modules: 1) pre-processing, 2) feature selection, and
accomplishing their tasks. In support of above statement, 3) Classification.
nowadays use of computer technology has been an integral Human beings possess natural ability to recognize
part of life. Computers are being used in pyramids of hundreds of faces by their visual system and cognition. That
applications, which range from simple to complex problem makes them recognize familiar faces even after a long period
solving methods. Among such contributions face recognition of time. Building an intelligent system similar to human
technology has emerged as useful tool to recognize features of perception system is still an active area of research. The
faces through their inherent traits. And it has been one of the researchers have suggested numerous algorithms and
most researched areas in the field of pattern recognition and methodologies for recognizing a face in an effective and
computer vision. However, due to its wide use in multitude of efficient manner. For this purpose, they have focused on
applications such as in biometrics, information security, law detection and recognition of traits and features for individuals
enforcement access control, surveillance system and smart such as nose, eyes, mouth, face shape position, size, and
cards. But it possesses many challenges for researcher that beside relationship among traits and features. Furthermore,
needs to be addressed. Face an object depends on facial ongoing research in face recognition tries to develop such
expressions, which constitute meaningful features. For systems that could work well in an effective and efficient
instance, pose invariance, illuminations and aging which are manner in multitude of real-world applications. In addition,
potential areas that require further investigation over previous many scholars have proposed and concluded that the accuracy
work. The result of previous researches reveals that facial of face recognition can further be improved through the use of
expressions are changing with respect to aging; therefore, they 3D faces [6]. The rest of this paper is arranged as follows:
could not be permanently modeled in face recognition. The Section II describes challenges related to the face recognition.
face recognition problem can be categorized into two main Section III explains face recognition databases. Section IV
phases: 1) face verification and 2) face identification. For elaborates methods and techniques in face recognition.
example, in real time system, face verification identifies the Section V illustrates widely used applications in face

42 | P a g e
www.ijacsa.thesai.org
 (IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 9, No. 6, 2018

recognition. Finally, Section VI gives conclusion and future can be divided into two kinds i.e. multi-view face recognition
directions. and face recognition across pose. Multi-view face recognition
can be considered as an annexure of frontal face recognition in
II. CHALLENGING AREAS IN FACE RECOGNITION which gallery image of every pose is considered. On the other
A. Aging hand, across a pose in face recognition, yield face with a pose
which has never been exposed before to a recognition system.
Aging is an inevitable natural process during the lifetime A good face recognition approach should provide good pose
of a person as compared to other facial variations. Aging tolerance and capability to recognize different poses. Several
effect can be observed under main three unique issues in this regard are still open such as lack of perceptive
characteristics: subspace pose variant images. And many of research have
1) The aging is uncontrollable: It cannot be advanced or been devoted to deal with this issue [9]-[13]. However, none
even delayed and it is slow and irreversible. of them has achieved 100% accuracy yet. There are some
other methods and approaches that are being used to tackle
2) Personalized aging Signs: Every human passes through
similar problem of face recognition. Furthermore variance and
different aging patterns. And these rely on his or her genes and changes in pose can be divided into three classes, namely:
many other factors, such as health, food, region, and weather 1) general algorithms, 2) two dimensional methods for face
conditions. recognition, and 3) three dimensional models [14].
3) The aging signs depend on time: The face of person at
a specific age will affect all older faces, but unaffected in D. Illuminations
younger age. Illumination is an observable property and effect of light.
It may also refer to lightning effect or the use of light sources.
B. Partial Occlusion Global illuminations are algorithms which have been used in
Occlusion refers to natural or artificial obstacles in an 3D computer graphics. Illumination variation also badly
image. It can be a local region of the face along with different affects the face recognition system. Thus it has been turned an
objects such as sunglasses, scarf, hands, and hair. They are area of attention for many researchers. However, it becomes
generally called partial occlusions. Partial occlusions tedious task to recognize one or more persons from still or
correspond to any occluding object. And the occlusion less video images. But it can be quite easy to extract desired
than 50% of the face is considered to be a partial occlusion. information from images when they are taken under a
The approaches to face recognition with partial occlusion are controlled environment along with uniform background. Also
classified into following three categories: 1) Part Based there are three methods that can be implemented to deal with
Methods, 2) Feature based methods and 3) Fractal-Based illumination problem. They are gradient, gray level and face
Methods [4]. Many areas of image processing have been reflection field estimation techniques. Gray level
impacted by partial occlusion such as recognition by ear is transformation technique carries out in-depth mapping with a
occluded due to earrings. Occlusion affects the performance of non-linear or linear function. Gradient extraction approaches
a system when people deceive it either by the use of are used to extract edges of an image in gray level. As
sunglasses, scarves, veil or by placing mobile phones or hands illumination is a factor that heavily affects the performance of
in front of faces. In some cases, other factors like shadows due recognition system obtained via face images or videos. These
to extreme illumination also act as occluding factors. Further, techniques are developed to suppress the effect of illumination
local approaches are used to deal with the problem of partially [9]-[13].
occluded faces which divide the faces into different parts [5].
However, this problem can be overcome by eliminating some III. FACE RECOGNITION DATABASES
of the features which create trouble while accurate recognition This database of faces was previously called The ORL
in the image. Mostly local methods are based on feature Database of Faces and it has a set of face images taken at the
analysis, in which best possible features are detected and then AT & T lab. This database was used for face recognition
they are combined. Another approach that can be applied for project which was carried out with the support of Speech,
this purpose is near holistic approach in which occlude Vision and Robotics Group of the Cambridge University,
features, traits and characters are eradicated and rest of the Department of Engineering. It includes 10 different images
face is used as valuable information. Different techniques are each having 40 diverse subjects. However, for some subjects,
being developed by the researchers to cope up with this the images were acquired under various conditions, i.e.
problem [7], [8]. variable light, facial expressions: smiling/sad, open/closed
eyes and facial details (glasses/without glasses). These images
C. Pose Invariance
were obtained with a dark consistent background having
Pose variance is yet another hurdle in achieving a position in upright, frontal. Details of the databases used in
successful face recognition system. People pose differently face recognition are described as under.
every time they take a picture. There is no standardized rule
for taking a pose. Therefore, it makes more difficult to A. AR Database
distinguish and recognize the faces from images with varying AR database was created by Computer Vision Center
poses. Pose variations degrade the performance of the facial (CVC), University of Alabama at Birmingham. It comprises of
features. In addition, many systems work under inflexible over 4000 colour images of 126 people’s faces. And they are
imaging conditions and as a result it affects the quality of divided into 70, 56 man women, respectively. Images feature
gallery images. The methods dealing with variation in pose frontal view faces with different facial expressions,

43 | P a g e
www.ijacsa.thesai.org
 (IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 9, No. 6, 2018

illumination conditions and occlusions (sun glasses, hair styles Yale face database is available in two volumes Yale face A
and scarves). The images of a single person were collected on known as yalefaces and Extended Yale face database B. In
two different days with a difference of 14 days. This database this database there are 15 different subjects (14 males and 01
is available online and can be accessed without any cost for female). These comprises of different conditions in facial
research and academic purposes. images such as variations in an expression like sad or normal
and happy, etc. This also depends upon other lighting
B. FERET Database conditions which consists of left, right or center light, and
The FERET database is being used in facial recognition picture having glasses and non-glasses were included.
for system evaluation. The Face Recognition Technology Extended Yale face database is a dataset of 2414 images of 38
(FERET) program is executed by joint collaboration between subjects. No variation in expression and no occlusions are
the Defense Advanced Research Projects Agency (DARPA) found in the images but more focus is on extracting feature apt
and the National Institute of Standards and Technology to illumination and they are available in cropped version.
(NIST). DARPA released a high-resolution, 24-bit color
version of these images in 2003. And it was tested over 2,413
still face images, representing 856 individuals. However, the
main motive behind development of FERET database was to
facilitate algorithm development and evaluation. Thus
initially, it requires a common database of facial images in
order to develop and test for the purpose evaluation. After
that, complications in image mentioned by the images should
enhance.
C. LFW Database
Labeled Faces in the Wild (LFW) is a database of face
photographs which was mainly developed for the
comprehension of unconstrained face recognition problem.
The data set has over 13,000 images of faces obtained via
web. And each face is labeled with the name of the person
whose picture was captured. However, roughly 1680 of the
pictured people contained two or more distinct photos in the
data set. However, the main constraint on these face images is
that they were detected by the Viola Jones face detector. They
are organized into four sets of LFW images, in which one is Fig. 1. Preprocessing steps of face recognition.
an original and other three are different types of aligned
images. The aligned images have "funneled images" LFW-a, IV. METHODS AND TECHNIQUES OF FACE RECOGNITION
which includes an unpublished method of alignment, and
"deep funneled" images. Among them, LFW-a, and the deep A. Eigen Faces
funneled images provide higher results over most of the face The word eigenface coined by German ―Eigen wert‖ The
verification algorithms for original images and funneled ―Eigen‖ literally mean characteristic and ―wert‖ mean value.
images. Eigenface is well establish algorithms that was used to
D. YouTube Face Database (YTF) recognize a feature in a face image. It is based on Principle
Component Analysis (PCA) [14]. In this method the
YTF database consists of face videos which were fundamental concept is to recognize the face by taking its
developed for unconstrained face recognition. In this database unique information about the face in question. Then encode it
shortest and longest clips are 48 frames and 6,070 frames to compare with the decode result of previously taken image
respectively. And the average length of a video clip is 181.3 as shown in Fig. 2. In eigenface method, decoding is
frames. And all the videos were taken from YouTube. An performed with the calculation of eigenvector and then it is
average of 2.15 videos is available for different subjects. represented as a matrix. However, Eigenface based face
E. Yale Database recognition systems is only suitable for images having the
frontal faces but some researches identify a face with different
The Yale Face Database includes 165 grayscale images in
poses have also been made [1]. Analyzing different results
Graphical interchange Format (GIF) of 15 individuals. They
drawn from the researchers the accuracy ratio has been much
are divided into 11 images/subject, each having different
improved in recent years as compared previous results. It is
facial expression or configuration: center-light, with/glasses,
expected to have an effective and efficient output in upcoming
no glasses, happy, sad, sleepy, normal, surprised, and wink.
years. A comparative study of the analysis results obtained by
Left-light, right-light.
different researchers by applying face recognition techniques
on the basis of Eigen Faces is given in Table I.

44 | P a g e
www.ijacsa.thesai.org
 (IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 9, No. 6, 2018

TABLE I. COMPARATIVE STUDY OF FACE RECOGNITION TECHNIQUES TABLE II. COMPARATIVE STUDY OF FACE RECOGNITION METHODS
BASED ON PCA USING ARTIFICIAL NEURAL NETWORS(ANN)

S Ye
Y Database Technique Accuracy Reference
S Techniq Accu # ar
ea Database Reference
# ue racy NN Based
r 20 IIT-Dehli 88.25% to Raja, A. S. et al.
1 SOM for Face
12 Database 98.3% [31]
20 70.0 Slavković et al. recognition
1 ORL Faces PCA
12 % [20] 20 BPC and RBC 96.66% & Nandini, M. et al.
2 -
13 Network 98.88% [32]
20 100.0 20
2 Face94 PCA Abdullah et al. [33] 3 Deep ID 3 99.53% Yi Sun, et al. [19]
12 % 15
20 Haoxiang Li et al.
20 Eigen 96.0 Saha, Rajib et al. 4 AFLW 99.00%
3 FRAV Face DB 15 [29]
13 Face % [51] 20 Multi PIE JunhoYim et al.
PCA Rahman, 5 CPF 99.50%
20 70.0 15 dataset [52]
4 - Eigen ArmanadurniAbd, 20 Sachin Sudhakar
14 % 6 AFLW 90.00%
Faces et al. [21] 15 Farfade [30]
92%
20 MuzammilAbdulra
5 Yale Database PCA to
14 hman et al. [37]
93%
2
Johannes Reschke
6 01 AT & T PCA
et al. [36]
4
Computer Vision
20 93.6 Md. Al-Amin
7 Research Projects PCA
16 % Bhuiyan [34]
dataset
20 PCA + 93.0 Jacky Efendi et al.
8 EmguCV library
17 RMF % [35]
2
Riddhi A. & S.M.
9 01 Yale Database PCA 98.18
Shah [46]
7

B. Artificial Neural Networks (ANN) Fig. 2. Face recognition using neural network.
ANN provides an effective feature recognition technique,
and it has been widely used after emergence of Artificial
Intelligence. This consists of network, where neurons are
arranged in the form of layers. Accuracy of face recognition
has been boosted with the aid of better deep network
architectures and supervisory methods. And recently few
remarkable face representation learning techniques are
evolved [15]. Using these techniques, deep learning (Fig. 3)
has got much closer to human performance. For evaluation
LFW face verification dataset has been used on tightly
cropped face images [15]. However, the learned face
representation could also add significant intrapersonal
variations. One of the most viable feature of Neural Networks
is it lessens the complexity. It learns from the training samples
and then works fine on the images with changes in lighting
conditions and increases accuracy [1]. The main drawback of Fig. 3. Face recognition by using SVM [38].
the neural network is a more time is needed for its training.
Initially Training is precursor step to get the desired results C. Support Vector Machine (SVM)
from the system as user point of view. SVM is the kind of supervised learning algorithm that uses
data for classification and regression analysis. SVM provides
After feature extraction, classifiers for face recognition advantages of being effective in high dimensions. SVM can be
such as the Radial Basis Function and Feed Forward Neural implemented to recognize the faces after facial feature
Network (FFNN) are the implemented. Moreover, study extraction [3]. SVM can yield better outcomes when the large
reveals that ANNs achieves improvement over face quantity of data set is selected directly with training (Fig. 2).
recognition [2]. The following comparative study in Table II However, Least Square Support Vector Machine (LS-SVM)
shows an accuracy ratio obtained through the use of ANNs. [50], [51] is among the popular one in SVM types that is being
successfully utilized for face recognition task.

45 | P a g e
www.ijacsa.thesai.org
 (IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 9, No. 6, 2018

This provides advantage of fast computation, speed along The main advantages of Gabor Wavelets Transform are
with high recognition rate [1]. Component-based SVM face feature reduction and its global feature representation in
classifier [16] is another variant of SVM in face recognition. face recognition [21], [29], [30]. Table IV shows comparative
The Support Vector Machine (SVM) classifier is a most analysis of Gabor filters by different researchers.
widely used technique that is being implemented on a wide
range of classification problems. Mostly these problems are in TABLE IV. COMPARATIVE STUDY OF FACE RECOGNITOIN METHODS
high dimensions and they are not linearly separable. SVM is BASED ON GABOR WAVELETS
useful in the advent of dealing with very high dimensional S# Year Database Technique Accuracy Reference
data. Researchers worked on SVM for classification of face FRGC & Ming et al.
1 2012 3D GPSR 95.80%
recognition and got better results as shown in the below CASIA [44]
Table III. 98.30%
PCA, 97.30%
TABLE III. COMPARATIVE STUDY OF FACE RECOGNITION METHODS Hyunjong Cho
2 2014 Yale Face DB LGBPHS
BASED ON SVM 99.20% et al. [43]
& DPL
S Ye Accu 99.70%
Database Technique Reference
# ar racy Dong Yi et al.
Least 3 2014 RBM PCA 99.5%
ORL Face [42]
1 2009 Square 96% Xie, Jianhong et al. [25]
Database Magnitude
SVM ORL
Phase of Bellakhdhar et
20 ORL Face 4 2013 Database, 99.90%
2 ICA, SVM 96% Kong, Rui et al. [26] Gabor, al. [45]
11 Database FRGCv2
PCA, SVM
2D- 93.8% Nooshin et al.
FERET 5 2015 T1-w dataset SVM RBF
Principal +0.3+- [38]
20 Database, 95.10 Le, Thai Hoang et al.
3 Component
11 AT&T % [27] MAHCOS Vinay.A et al.
Analysis, 6 2015 ORL 97.50%
Database Distance [39]
SVM
Shervin
20 97.78 Bhaskar Anand & IIT-Dehli
4 Yale Faces SVM 7 2015 PCA 99.20% Minaee et al.
16 % Prashant K Shah [24] Database
[40]
D. Gabor Wavelet FERET & Leonardo A.
8 2015 92.80%
CMU-PIE Camenta [41]
Dennis Gabor in 1946 introduced a tool for signal
processing in noise removal and named as Gabor filter. Gabor E. Hidden Markov Models
wavelets technique is being widely used for face tracking and Hiddel Markov Model (HMM) is another statistical
position estimation in face recognition. While an image modelling technique in which the system undergoes in
representation using the Gabor wavelet transform provides Markov process with hidden states (Fig. 5). This model was
both the spatial relations and spatial frequency structure. As proposed in 1960 and provided a significant contribution
shown in Fig. 4, it has a characteristic that allow it to delineate towards speech recognition. HMM is well established method
the properties of spatial localization, spatial frequency in reinforcement learning, temporal pattern recognition, and
selectivity, and orientation [17], [18]. Gabor Wavelets works bioinformatics applications. Currently it is being implemented
well over extraction of edge and shape information and it to recognize face expressions. Also it can be applied to the
represents the faces in a compact way which is more similar to video sequences for face recognition. It needs a sequence of
the feature based methods [19]. 1D and 2D images for experimental purposes [21]; but firstly
these images should be converted into a chronological
sequence of 1D or spatial. However, model consists of two
processes, in which first Markov Chain process having a finite
number of states is not viewed explicitly. While in other
process each state constitutes a set of probability density
function connected with it [1]. Although for research,
generally 5-state HMM is designed for face recognition
system. 5-state HMM is grouped into five facial features such
as eyes, nose, mouth, chin and forehead for frontal view face
images as shown in Fig. 6 [47]. But the number of states can
be added or removed which depends upon system’s
requirement.

Fig. 4. Gabor wavelet process of recognition.

Fig. 5. Three states of transition from left to right for HMM [25].

46 | P a g e
www.ijacsa.thesai.org
 (IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 9, No. 6, 2018

and is recognized. In this case, only legitimate persons are

allowed to access account.
B. Security
Security is a most important precursor at all places.
Computer security is being carried out with use of face
recognition application. In this regard, image database is being
used for investigation purposes [1]; for instance, searching
image for authentication of licensed drivers to search missing
peoples, immigrants in law enforcement agencies, General
identity verification [1], Electoral registration, banking,
electronic commerce, identifying newborns, national IDs,
passports, employee IDs.
C. Surveillance
Fig. 6. Hidden Markov Model process of recognition. The word surveillance has been derived from
a French phrase which means "watching over". Here
In other case, 7-State HMM [48] provide more details (sur means "from above‖ and veiller means "to watch").
which boosts the performance of the face recognition system. Surveillance is used to monitor the individual’s behavior,
Various researchers have worked to get satisfactory outcomes activities, or beside other related information for ensuring the
by applying different algorithms in this model. However, a people safety. This can be achieved by means of electronic
new enhanced model The Adaptive Hidden Markov Model equipment i.e. closed-circuit television (CCTV) cameras) or
(AHMM) [49] is proposed by authors to find out the issues of interception of electronically transmitted information.
identifying the faces via a video sequence. The comparative Surveillance system offer no. of benefits to different
result in Table V depics the accuracy ratio of HMM. organizations. For example, it is being used by governments
for intelligence gathering, control the crime, monitoring the
TABLE V. COMPARATIVE STUDY OF FACE RECOGNITION METHODS process, person, crowd or object, or the inquiry of crime.
BASED ON HIDDEN MARKOV MODELS (HMM) However, on the other side, surveillance is often considered as
S# Year Database Technique Accuracy Reference a violation of privacy, and in such cases it is often criticized
by civil societies, groups and activists. Liberal
ORL Sub- Sharif, democracies have laws which bounds local governments and
95.25% &
1 2013 FaceDB, Yale Holistic Muhammad et
Face DB HMM
94.45%
al. [28] law agencies to use surveillance, usually restricting them in
those circumstances where public safety is
CK+ UNBC- MIL - Chongliang compromised. Legitimate organizations have often been
2 2015 85.23%
McMaster HMM Wu et al. [22]
imposed such domestic restrictions. However,
Samiksha international surveillance is similar among all types of
3 2015 UMIST MS-HMM 93.66% Agarwal et al. countries. Moreover researchers are trying to achieve more
[23]
improved and contented results through the use of latest
MSR- DMM- Chen Chen et algorithms in face recognition.
4 2013 90.50%
Action3D HMM al. [24]
D. Time & Attendance
James Lyons
5 2015 SCOP HHblits 93.80%
et al. [25] Biometric Time Attendance technologies have been used
for Access Control solutions and these are among the latest
V. APPLICATIONS OF FACE RECOGNITION solutions over traditional systems [23]. In this technology,
users are required to expose their face into the machine’s
There are many applications where face recognition
camera by making a certain distance and remove any physical
techniques are successfully used to perform a specific task.
contact with the device. This eliminates any possibility of
Few of them are described as under:
being tempering or machinery alteration through its non-
A. Access Control contact method procedure. Face Recognition system captures
Access control allows the authorized group of users to specific features from a human face and record it in the form
access the personal account by logon through their email of mathematical template as depicted in Fig. 7. In order to
account using computer accessing bank account through ATM recognize the face, facial image is normalized as to line-up
machine. But using face recognition system face pictures are eyes and mouth. Then it performs matching with mathematical
taken under natural conditions such as frontal face images. vectors from database. Finally face recognition system verifies
Such kind of systems yields optimal accuracy without any face and allows for marking attendance or access transaction.
intervention from the user. These automatic face recognition These machines could also be implemented for other
systems are also used to view and control a user activity on PC solutions, where biometric identification/verification is
or ATM machine; for example, when users leave the PC required; such as canteen management, salary distribution, and
without proper closing their files and folders for a social services.
predetermined time. Then system halts until user again logon

47 | P a g e
www.ijacsa.thesai.org
 (IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 9, No. 6, 2018

 The development trends and achievements in the realm

of face recognition shows that a lot of researchers have
been carried out in last four decades.
 Currently, face recognition system has been
implemented for many real-time applications, but still
it suffers from several challenges that need to be
addressed in order to design a well-established face
recognition system.
 Developed face recognition techniques could be
analyzed over varying facial expression i.e. under
varying lighting conditions and pose. And evaluation
could be performed using benchmark and latest face
databases.
 Similarly to the face image recognition, the video
image recognition is more complicated that needs to be
researched.

Fig. 7. Facial recognition.

Further, it is suggested that for recognition of video
images, YouTube Faces could be analyzed for evaluation.
E. Pervasive Computing Furthermore, recognition of emotional human behavior has
emerged recently as a promising [52] research area for
The aim of pervasive computing is to create a sensor based scholars that should be exploited in future. Finally, it is
network as to make smart devices. Hence, sensor network is concluded that still there remains a gap in terms of study in
used to collect, process and send data, and eventually, it can face recognition system that requires to be filled in order to
understand its surroundings and improves the human improve its accuracy and efficiency.
capability and quality of life. However, pervasive computing
uses wireless communication and networking technologies, REFERENCES
mobile devices, wearable computers, embedded systems, [1] Muhammad Sharif et al.: ―Face Recognition: A Survey‖, Journal of
Radio Frequency Identification Devices (RFID) tags, Engineering Science and Technology Review 10 (2) (2017) 166- 177
middleware and software agents. Pervasive computing is [2] Lacey et al.: ―A Longitudinal Study of AutomaticFace Recognition‖,
IEEE ICB, 2015 pp: 1- 8
being widely used in number of applications, for instance in
energy, consumer, healthcare, production, military, safety, and [3] Ms. Snehal Houshiram Gorde1, et al.‖ A Review on Face Recognition
Algorithms‖ Volume III, Issue I Issn No.:2350-1146, I.F-2.71
logistics.
[4] Azeem, Aisha, et al. "A survey: face recognitiontechniques under partial
One of the examples of pervasive computing is a smart occlusion." Int. Arab J. Inf. Technol. 11.1 (2014): 1-10.
Watch developed by Apple Watch. It informs a user for [5] Tarrés, Francesc, Antonio Rama, and L. Torres. "Anovel method for
incoming phone call and allows him to complete the call using face recognition under partial occlusion or facial expression variations."
Proc. 47th Int’l Symp. ELMAR. 2005.
watch [1].
[6] Sharif M., Mohsin S., Hanan R., Javed M. and Raza M., "3D Face
Recognition using Horizontal and verticalMarked Strips", Sindh
VI. CONCLUSION AND FUTURE DIRECTIONS University Research Journal (SURJ), 43(01-A), (2011)
Study of face recognition has remained a striving area for [7] Jia, Hongjun, and Aleix M. Martinez. "Face recognitionwith occlusions
researchers for many years. In this paper, a comprehensive in the training and testing sets." AutomaticFace & Gesture Recognition,
study was performed over different face recognition methods. 2008. FG'08. 8th IEEEInternational Conference on. IEEE, 2008.
After detailed analysis it revealed that PCA is best suited [8] Zhou, Zihan, et al. "Face recognition with contiguousocclusion using
markov random fields." Computer Vision, 2009 IEEE 12 th International
technique when dimension of features is higher for original Conference on. IEEE, 2009.
face images, whereas eigen faces image features method work
[9] Huang, Fu Jie, et al. "Pose invariant face recognition." Automatic Face
well for frontal face recognition. Among face recognition and Gesture Recognition, 2000. Proceedings. Fourth IEEE International
methods, the most popular are Neural Networks, Support Conference on. IEEE, 2000.
Vector Machine, Sparse Representation based Classification [10] Chai, Xiujuan, Shiguang Shan, and Wen Gao. "Pose normalization for
(SRC), Linear Regression Classification (LRC), Regularized robust face recognition based on statistical affine transformation."
Robust Coding (RRC) and Nearest Feature Line ((NFL). Information,Communications and Signal Processing, 2003 and
These methods provide better results when the image FourthPacific Rim Conferenceon Multimedia. Proceedings of the2003
Joint Conference of the Fourth InternationalConference on. Vol. 3.
dimension is under 150 or more. Furthermore, it is suggested IEEE, 2003.
that PCA, SVM, NN and Eigen methods still need to be [11] Wright, John, and Gang Hua. "Implicit elastic matching with random
researched so that more satisfactory results could be achieved projections for pose-variant facerecognition." Computer Vision and
for face recognition. Moreover, in this paper we also Pattern Recognition, 2009. CVPR 2009. IEEE Conference on. IEEE,
mentioned state of the art face recognition image database and 2009.
face technology benefits in various applications. However, [12] Zhang, Wuming, et al. "3D aided face ecognitionacrosspose
main findings of this research are highlighted as under: variations."Biometric Recognition. Springer Berlin Heidelberg, 2012.
58-66.

48 | P a g e
www.ijacsa.thesai.org
 (IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 9, No. 6, 2018

[13] Shah J. H., Sharif M., Raza M. and Azeem A.," Facerecognition across [34] Shah J. H., Sharif M., Raza M. and Azeem A.," Facerecognition across
pose variation and 3S problem", In TÜBİTAK Academic Journals pose variation and 3S problem", In TÜBİTAK Academic Journals
(2012) (2012)
[14] Zhang, Xiaozheng, and YongshengGao. "Facerecognition across pose: [35] Zhang, Xiaozheng, and YongshengGao. "Face recognition across pose:
A review." Pattern Recognition42.11 (2009): 2876-2896. A review." Pattern Recognition 42.11 (2009): 2876-2896.
[15] Wang, Jizeng, and Hongmei Yang. "Face detectionbased on template [36] Azeem, Aisha, et al. "A survey: face recognition techniques under
matching and 2DPCA algorithm." Imageand Signal Processing, 2008. partial occlusion." Int. Arab J. Inf. Technol. 11.1 (2014): 1-10.
CISP'08. Congress on. Vol. 4. IEEE, 2008. [37] Chai, Xiujuan, Shiguang Shan, and Wen Gao. "Posenormalization for
[16] Huang, Jennifer, Volker Blanz, and Bernd Heisele."Face recognition robust face recognition based on statistical affine transformation."
using component-based SVMclassification and morphable models." Information, Communications and Signal Processing, 2003 and Fourth
Pattern Recognitionwith Support Vector Machines. Springer Berlin Pacific Rim Conference on Multimedia. Proceedings of the 2003 Joint
Heidelberg,2002.334-34 Conference of the Fourth International Conference on. Vol. 3. IEEE,
[17] Jin, Yi, and Qiu Qi Ruan. "Face recognition using gabor-based improved 2003.
supervised locality preservingprojections." Computing and Informatics [38] Wei, Xingjie, Chang-Tsun Li, and Yongjian Hu. "Robustface
28.1 (2012): 81-95. recognition under varying illumination and occlusion considering
[18] Bellakhdhar, Faten, Kais Loukil, and Mohamed Abid."Face recognition structured sparsity." Digital Image Computing Techniques and
approach using Gabor Wavelets, PCA and SVM." IJCSI International Applications (DICTA), 2012 International Conference on. IEEE, 2012.
Journal of Computer Science Issues 10.2 (2013): 201-206. [39] Zhao, Wenyi, and Rama Chellappa. "Illumination-sensitive face
[19] Kar, Arindam, et al. "Classification of high-energizedgabor responses recognition using symmetric shape-from-shading."Computer Vision and
using bayesian PCA for human face recognition." 2.2(09) Pattern Recognition, 2000. Proceedings. IEEE Conference on. Vol. 1.
IEEE, 2000.
[20] Dong Yi, Zhen Lei, Shengcai Liao and Stan Z. Li
"SharedRepresentation Learning for Heterogeneous FaceRecognition‖ [40] Sharif M., Mohsin S., Jamal M. J. and Raza M.,"Illumination
2015 pp: 1-15 Normalization Preprocessing for face recognition", IEEE International
Conference on Environmental Science and Information
[21] Chongliang Wu1, Shangfei Wang _1 and Qiang Ji"Multi-Instance ApplicationTechnology (ESIAT), , 44-47 (2010)
Hidden Markov Model For Facial Expression Recognition 2015 IEEE
[41] Sharif M., Raza M. and Mohsin S., "Face RecognitionUsing Edge
[22] Samiksha Agrawal, ―Facial Expression Detection Techniques: Based on
Information and DCT", Sindh Univ. Res. Jour. (Sci. Ser.) Vol.43 (2)
Viola and Jones algorithm and Principal Component Analysis 2015 209-214, (2011)
IEEE
[42] Sharif M. and Saad A., "Enhanced SVD Based FaceRecognition",
[23] Chen Chen et al., ―Real-time human action recognitionbased on depth Journal of Applied Computer Science & Mathematics, Suceava, Vol.
motion maps‖ 2013 Springer 12(6) ,49-53,(2012)
[24] Bhaskar Anand & Prashant Face Recognition using SURF Features and [43] Cho, Hyunjong, Rodney Roberts, Bowon Jung, Okkyung Choi, and
SVM Classifier‖ ISSN 0975- 6450 Volume 8, Number 1 (2016) pp. 1-8
Seungbin Moon. "An efficient hybrid face recognition algorithm using
[25] Xie, Jianhong. "Face recognition based on Curvelettransform & LS- pca & gabor wavelets‖ 2014
SVM." Proceedings of the 2009 International Symposium on [44] Ming, Yue, QiuqiRuan, and Xueqiao Wang. "Efficient 3dface
Information Processing (ISIP’09), Huangshan, PR China. 2009.
recognition with Gabor patched spectral regression." Computing and
[26] Kong, Rui, and Bing Zhang. "A New Face Recognition Method Based Informatics 31.4 (2012): 779-803.
on Fast Least Squares Support Vector Machine." Physics Procedia 22 [45] Bellakhdhar, et al. "Face recognition approach using Gabor Wavelets,
(2011): 616-621.
PCA and SVM."IJCSI International Journal of Computer Science Issues
[27] Le, Thai Hoang, and Len Bui. "Face recognition based on SVM and 10.2 (2013): 201-206.
2DPCA."arXiv preprint arXiv:1110.5404 (2011).
[46] Riddhi A. Vyas1, Dr.S.M. Shah ―Comparision of PCA and LDA
[28] Sharif M., Ayub K., Sattar D. and RAZA M., "Real TimeFace Techniques for Face Recognition Feature BasedExtraction with
Detection", Sindh Univ. Res. Jour. (Sci. Ser.) Vol. 44(4), 597- 600, Accuracy Enhancement‖ 2017, IRJET pages:3332-3336
(2012)
[47] Nefian, Ara V., and Monson H. Hayes III. "Hidden markov models for
[29] KalavdekarPrakash, N. "Face Detection using NeuralNetwork." face recognition." choice 1 (1998): 6.
International Journal of Computer Applications(0975–8887) 1.14 [48] Miar-Naimi, H., and P. Davari. "A new fast and efficient HMM-based
(2010). face recognition system using a 7-state HMM
[30] Li, Yongmin, et al. "Multi-view face detection usingsupport vector [49] Liu, Xiaoming, and Tsuhan Chen. "Video-based face recognition using
machines and eigenspacemodelling." Knowledge-Based Intelligent adaptive hidden markov models." Computer Vision and Pattern
Engineering Systems and Allied Technologies, 2000. Proceedings. Recognition, 2003. Proceedings. 2003 IEEE Computer Society
Fourth International Conference on. Vol. 1. IEEE along with SVD coefficients." (2008).
[31] Raja, A. S., and V. JosephRaj. "Neural network basedsupervised self- [50] Xie, Jianhong. "Face recognition based on Curvelet transform and LS-
organizing maps for face recognition." International Journal on Soft SVM." Proceedings of the 2009 (ISIP’09)Huangshan,
Computing 3.3 (2012).
[51] Zhang, Xinming, and JianZou. "Face recognition based on sub-image
[32] Nandini, M., P. Bhargavi, and G. Raja Sekhar. "FaceRecognition Using feature extraction and LS-SVM." Computer Science and Software
Neural Networks." International JournalOf Scientific and Research Engineering, 2008 International Conference on. Vol. 1. IEEE, 2008.
Publications 3.3 (2013): 1. Conference on. Vol. 1. IEEE, 2000.
[52] Maha Jazouli, Aicha Majda, Arsalane Zarghili. "A $PRecognizer for
[33] Abdullah, Manal, MajdaWazzan, and Sahar Bo-Saeed."Optimizing Face
Automatic Facial Emotion Recognition using Kinect Sensor." 978-1-
Recognition Using PCA." arXiv preprintarXiv:1206.1515 (2012).
5090-4062-9/17/$31.00 ©2017 IEEE.

49 | P a g e
www.ijacsa.thesai.org