effective sharing and testing of data on web services ... · work combines persuasive cued click...
Post on 18-Jun-2018
212 Views
Preview:
TRANSCRIPT
-
National Conference on Green Computing Trends in Information and Communication Technology
(NCGCTICT-2015)
ISSN: 2349 - 641X www.internationaljournalssrg.org Page 1
Effective Sharing and Testing of Data on Web Services using
Combinatorial Testing
R.DHANUSREE ME-II Year
Computer Science And Engineering
Arunai College Of Engineering
Thiruvannamalai
S.MOHANARANGAN Assisstant Professor
Computer Science And Engineering
Arunai College Of Engineering
Thiruvannamalai
Abstract- Web injection attack exploits web application security by inserting malicious script into the web
application. The need for this kind of attack is to gain
access to application data or database. Here the focus is on
securing web application against web injection attacks. To
detect injection vulnerabilities in web services, we use
combinatorial interaction testing the main focus is on
Cross-Site Scripting Attack (XSS). By analyzing the
behavior of the Web service the model can detect the
presence of the injection vulnerabilities. This proposed
work combines persuasive cued click points and password
guessing resistant protocol to reduce the guessing attacks
as well as motivate users to select more random and
difficult passwords to guess. An algorithm for anonymous
sharing of private data among parties is been developed.
Assigning ID numbers to nodes ranging from 1 to n, the
identity received are unknown to the other member. Using
serial number assigned the complex data can be shared.
The new algorithms are built on top of a secure sum
Security operation using service discovery identities and
ids vulnerability detection secured protocol.
Keywords- Cross-site Scripting Attack, Persuasive cued click point, Password guessing resistant protocol and
Combinatorial Interaction Testing.
I. INTRODUCTION
Web service is a systemized way of combining Web-based
applications using XML, SOAP, WSDL, UDDI open
Standards over an Internet protocol. XML is used to tag the
data; SOAP is used to transfer the data between web
applications. WSDL is used for describing the services that
are available and UDDI is used for listing what services are
available in the application. Web service used for business purpose to communicate with the client. By using Web
services the organizations can exchange data without
knowledge of others in IT systems. Applications in Web
services from different sources can communicate with each
other in less time, because everything is in the form of XML. Web services are not fixed to single operating system or programming language. For example, Java can work with
Perl. Windows applications can work with UNIX applications. Web application is the foundation for each
activity in the internet, through which all the informations are
available in the internet. This demand for web application also
makes the attackers to exploit the vulnerabilities.
Web Injections Browser behavior is exploited by
constructing malicious input strings using input validation as
the form of application vulnerabilities, malicious purpose varies depending on the type of injection; all malicious
activities are used to compromise the three region of
information security: confidentiality, integrity, and
availability. In injection techniques the HTTP headers are
used to pass input data to the server-side web application, the
GET and POST methods will include malicious parameters
processed by the web application.
Many type of malware has the ability to insert malicious code into the client side browser, without the knowledge of
the server-side application which is undetected normally. The
aim of the attacker is to obtain secret data by performing
illegal activities, he identifies the system which is vulnerable
and using that system the attacker will gain information about
the victim. Multiple vulnerabilities possible in dynamic web
applications, Cross-site scripting is one among the top ten
vulnerabilities according to Open Web Application Security
Project (OWASP). XSS vulnerabilities have been introduced
since 1990s. Many sites are affected by this attack are Twitter,
Face book, MySpace, YouTube and Orkut. Nowadays the cross-site scripting attack exceed buffer overflows to become
the most commonly occurring security vulnerability, Some
researchers in 2007 found that 68% of websites are open to
XSS attacks.
This paper will focus on how the web application is
secured against web injection attacks, with the aim to develop
an understanding of how web injection attacks may be
detected and ultimately prevented. Based on the findings the project will aim to develop an appropriate web injection
solution.
Organization This paper is organized as follows. In
Section 2 we present the security problems in the presence of
Cross-site scripting attack. In Section 3 we present proposed
approaches we use to identify the presence of vulnerabilities
and use some algorithms to provide authentication against the attacks. In Section 4 we present Implementation part of our paper how the algorithm works and provide
authentication against attacks. In Section 5 we present the
http://www.internationaljournalssrg.org/ncgctict-2015.htmlwww.internationaljournalssrg.orghttp://www.webopedia.com/TERM/A/application.htmlhttp://www.webopedia.com/TERM/O/open.htmlhttp://www.webopedia.com/TERM/S/standard.htmlhttp://www.webopedia.com/TERM/P/protocol.htmlhttp://www.webopedia.com/TERM/T/tag.htmlhttp://www.webopedia.com/TERM/O/operating_system.htmlhttp://www.webopedia.com/TERM/P/programming_language.htmlhttp://www.webopedia.com/TERM/J/Java.htmlhttp://www.webopedia.com/TERM/P/Perl.htmlhttp://www.webopedia.com/TERM/M/Microsoft_Windows.htmlhttp://www.webopedia.com/TERM/U/UNIX.html
-
National Conference on Green Computing Trends in Information and Communication Technology
(NCGCTICT-2015)
ISSN: 2349 - 641X www.internationaljournalssrg.org Page 2
Security analysis, here the resulting outcome is analyzed with
the concept.
II. CROSS-SITE SCRIPTING ATTACK
Around 80% of all web application vulnerabilities, Cross-
Site Scripting (XSS) is one of the most predominant web
application injection attacks and attached to other injection
vulnerabilities. XSS mainly occurs due to inadequate input
filtering procedures utilized by the web application host. Like
most web injection attacks, successful XSS exploits may lead
to compromised authentication information, privilege risk and
possible revelation of confidential information. As a result of
vulnerabilities on the server-side of the application, this type
of attack is achieved. Using the client-side web browser,
complexity is further added to the detection and gathering evidence of a successful attack. This causes victim browser to
execute JavaScript crafted by the attacker to gain access rights
to sensitive data, session cookies, and other informations.
This attacks are made to steal confidentiality of sensitive data,
undetermined authorization schemes, defraud users and
defame web sites.
XSS allows a user to accidentally send malicious data to
him through that application. Attackers often perform XSS attack by crafting malicious URLs and tricks users to clicking
on malicious link.
The XSS occurs in the system using Script. Embedded
JavaScript has the ability to execute on the users browser
with same permission.
An attacker waits for their victim to view and execute the
injected code using scripting attacks.
Figure 1: Traditional XSS Web Application Hijack Scenario
If the software did not validate user input, a malicious user can add the malicious Comment within the tags.
When other users viewed the comment, it might look
something like a normal code. There are two basic techniques
to accomplish an XSS attack. The first technique is to store malicious code in database and when accessed by client will
be executed by the browser at the client side. The second
technique requires that the victim without the knowledge of
malicious link clicks on the link resulting in execution of
malicious code.
A.XSS Vulnerability causes:
An attacker could write malicious scripts; a JavaScript with an infinite loop could which makes the victims browser
unusable, forcing them to quit the browser. Similarly the
attacker could manipulate the window, by shrinking it, closing
it, or making it move randomly across the screen, or
manipulate the Document Object Model to embed or alter text
and images. A more sophisticated attack could use DOM
manipulation to alter from values as part of an attempt to
gather information intended for the vulnerable application.
The action could be switched to post the submitted data to
a logging script on the attackers site, for instance. Dom
manipulation via JavaScript would make this attack
mechanism highly difficult to detect.
An XSS attack could also use browser-specific
vulnerabilities in scripting implementations to scrape
information out of files on a users hard drive. Attackers target
is to obtain access to sensitive information stored on their systems, sending malicious URLs in email designed to appeal
to the specific intended victims. The most common behavior
of XSS attacks, however, is to gather cookies. Cookies are a
technology initially designed for Netscape Navigator 1.0 to
mitigate some of the problems stemming from HTMLs nature
as a stateless protocol. They are text files that reside on a
users computer and store name-value pairs along with some
metadata. Cookies are commonly used to store information
intended to be persistent during a browser session or from
session to session, such as session IDs, user preferences, or
login information.
III. OUR APPROACH
In this section we present some of the concept to provide
better authentication against injection vulnerabilities and
perform testing to measure the presence of vulnerabilities in web application using a better methodology.
A. Taxonomy of Authentication
The first stage in securing any computer system is to verify
the identity of users. This process of checking a users identity is referred to as user authentication.
Passwords are often used to authenticate any user.
Normally, user created passwords which are easy to remember
and recall. which is easily cracked by the third party with the
malicious intentions. So it is necessary to create passwords with more security measures. If the user creates any password
it has to be more secure.
http://www.internationaljournalssrg.org/ncgctict-2015.htmlwww.internationaljournalssrg.org
-
National Conference on Green Computing Trends in Information and Communication Technology
(NCGCTICT-2015)
ISSN: 2349 - 641X www.internationaljournalssrg.org Page 3
Most of the user create text based password which are
vulnerable for different class of attacks like brute force,
dictionary based attack, shoulder surfing attack. In spite of
such vulnerabilities users are tend to select short guessable
password. Unfortunately, such passwords can be easily
breakdown by third parties. To avoid this problem, the idea of
image based passwords is introduced. The graphical
passwords are more popular since they are better than Text
based passwords, because people are better at memorizing
graphical passwords than text-based passwords.
1. Persuasive Cued Click Points (PCCP)
The persuasive technology was first introduced by Fogg.
Persuasive feature is added to Cued-Click Point concept to
allow users to select less predictable passwords. To avoid hotspot problem the viewport is positioned randomly rather
than specifically. Using such information attackers can their
improve guesses and form new hotspots. Since all click-points
are hotspots, it makes it more difficult to select passwords.
This persuasive technology guides and encourages users to
select stronger passwords, rather than improving system-
generated passwords. Advantage of image based
authentication system using persuasive cued click point is
resistant to many security attacks such as dictionary attack,
brute force attack, hotspots, Guessing attacks, capture attacks,
etc., In this concept during image selection we select multiple click-points on the single image instead of selecting click-
points on a multiple image to reduce time usage.
Figure 2: Registration Phase
This project proposes a new protocol called Password
Guessing Resistant Protocol, designed to restrict Brute force
and dictionary attacks on password. PGRP limits the total
number of failed login attempts from unknown remote hosts,
legitimate users can make several failed login attempts before
being challenged with an ATT.
PGRP provides protection against some threats such as
key logger spy ware. Mouse is provided instead of keyboard
to enter our image password to protect our password from key
loggers. PGRP prevent password guessing attack without answering ATT challenges, since it is more effective. It also
provides convenient login experience, e.g., fewer ATT
challenges for legitimate users. PGRP appears for
organizations with large number of user accounts.
B. Homomorphism Vulnerability Algorithm
An algorithm for anonymous sharing of private data
among multiple parties has been developed. This technique
assigns ID numbers to each nodes ranging from 1 to N
iteratively. This assignment is not known to the members of
the other group. Resistance to collusion among other members
is verified in an information theoretic sense when private
communication channels are used. Using the serial numbers
assigned the complex data can be shared and has applications
to other problems in privacy preserving data mining, collision
avoidance in communications and distributed database access.
Without trusted central authority the required computations are distributed. Existing and new algorithms for assigning
anonymous IDs are examined with respect to trade-offs
between communication and computational requirements. The
new algorithms are built on top of a secure sum data mining
operation using service discovery identities and IDS
vulnerability detection secured protocol. A homomorphism
vulnerability algorithm for distributed solution of certain
polynomials over finite fields enhances the scalability of the
algorithms.
C. Combinatorial Interaction Testing
The popular selection proposed approach is combinatorial
interaction testing (CIT), where the developer selects a
strength t and then computes a covering array (a set of
configurations) in which all t-way combinations of
configuration option settings appear at least once. In prior work, we demonstrated several limitations of the CIT
approach. In particular, we found that a given systems
effective configuration space. We also found that effective
configuration space may not be well approximated by t-way
covering arrays. Based on these insights we have developed
an algorithm called interaction tree discovery (iTree).
Interaction tree discovery is an iterative learning algorithm
that efficiently searches for a small set of configurations that closely approximates a systems effective configuration space.
http://www.internationaljournalssrg.org/ncgctict-2015.htmlwww.internationaljournalssrg.org
-
National Conference on Green Computing Trends in Information and Communication Technology
(NCGCTICT-2015)
ISSN: 2349 - 641X www.internationaljournalssrg.org Page 4
Figure 3: Proposed System Model
The Proposed approach follows the methodology as the
above model starting from preparing test to analyzing the
output from the HTTP header. By analyzing the behavior of
the web service the model can detect the presence of the
injection vulnerabilities in the Web Services. Based on the gathered information we find the different way of input that
are being provided to the web service. We provide the
workload in a random manner as well as in a sequence with
the absence of any attack.
Figure 4: Proposed model on the basis of input
For generating XSS attack scripts we classify the large
data set of script that are capable of emulating XSS attacks
and classify them on the basis of length, 4 different types of encoding techniques namely URL encoding, Base64 encoding
,HTML encoding, Hex encoding.. This classification makes it
easier to generate the attack load for the input parameter.
The priority of attackload generation for testing each
input is varied for example for user controled input The
priority of testing a double quote script with length more than
75 is the first priority ,Followed by single quoted script and
followed by the different encoded scripts.The flowchart is
given below.
Figure 5: Flowchart
For other kind of input the priority is different for
generation and testing of the attackload.For detection of the
vulnerability we use the response header of the SOAP
message exchanged. Based on the results of HTTP status code
in the header of the SOAP message response, we determine
the existence of vulnerabilities in Web Services, described
below. If the header contains the code 200 OKAND the
server ran the SOAP message with the XSS attack, THEN
there is a Vulnerability Found (VF) in the Web Service.
OTHERWISE, if the SOAP message describes the existence of a syntax error or warning about the presence of an attack,
THEN there is No Vulnerability Found (NVF) in the Web
Service. If the header contains the code 400 Bad request
message, e.g. request format is invalid: missing required
soap: Body element, THEN there is No Vulnerability Found
(NVF) in the Web Service.
If in the absence of attacks, the header contains the code
500 Internal Server Error AND there was information disclosure in the SOAP message. AND if in the presence of
XSS attack, the header contains the code HTTP 200 OK,
THEN there is a Vulnerability Found (VF) in the Web
Service. Based on the behavior we can further classify on the
basis of response header if needed. In Any case there is no
response then the result remains inconclusive.
IV. IMPLEMENTATION
A. Welcome Page
The basic step for any authentication scheme is to become
a valid user of that system. We use Graphics or Image based
http://www.internationaljournalssrg.org/ncgctict-2015.htmlwww.internationaljournalssrg.org
-
National Conference on Green Computing Trends in Information and Communication Technology
(NCGCTICT-2015)
ISSN: 2349 - 641X www.internationaljournalssrg.org Page 5
authentication which is more reliable, for performing this kind
of authentication user has to create an account with the user id
and the password in textual based login system.
This step is necessary to have an entry in the administrator
level to check whether the user is authenticated or not. When
the new user arises, the user must select new user id, password and proceed to become a valid user
B. Registration Phase
During registration the system will ask the user to provide
information such as user id, password, mail id etc., the user
should provide the textual based passwords. The information
that is provided during registration is maintained by the
administrator to check whether the intended user is
authenticated user or not. After registering, whenever the user need to access any application the login page appears. To
conform as an authenticated user, the users specify user id and
password in a textual form.
C. PCCP Algorithm
After the textual based login system, now the user
entering to the PCCP algorithm step, here we use Image based
passwords along with the textual password via cued click-
points. During this phase user has to select how many click
points needed to create the password which will show the
strength of the password security. To improve the strength of
the security in the graphical passwords the number of click-
points in the image can be increased.
Figure 6: PCCP Algorithm
D. Automatic Turing Test
The image appears after that enter the authentication code and click the upload button, by clicking on the correct
pixel in the image the authentication code which we uploaded
will be generated.
Figure 7: Automatic Turing Test
After that automated Turing Test (ATT) is performed on
the authenticated code, if the user entered code and code generated during click-points in the image are compared. If
the authentication code is same then system enters to the
corresponding users page.
E. White List
With the use of the homomorphism vulnerability
algorithm the frequent failed login attempts made by any
system can be identified by their Id assigned using this
algorithm. At last system maintains a white list in which the
system which performed failed attempt frequently is listed and
further access is denied for those ID for further attempt.
V. SECURITY
A. Password Guessing Attack
The brute force attack and dictionary attacks are most
basic password guessing attack against PCCP. In Brute force
attack the attacker tries all possible code, combination, or
password to obtain the correct code by guessing, which is a
time consuming attack. Brute force attacks are avoided by
selecting complex password because it requires more time to
hijack the password. Dictionary Attack: Used by the attacker
to identify the users password by using dictionary of common
words.
B. Capture Attack
By intercepting the data entered by the user, the attacker
can directly obtain passwords or by tricking users to reveal
their passwords. Key loggers can be prevented by using
mouse rather than keyboard to enter graphical password using
PGRP protocol.
http://www.internationaljournalssrg.org/ncgctict-2015.htmlwww.internationaljournalssrg.org
-
National Conference on Green Computing Trends in Information and Communication Technology
(NCGCTICT-2015)
ISSN: 2349 - 641X www.internationaljournalssrg.org Page 6
C. Hotspots
The area which is chosen often by the user in an image as
a part of password is referred to as Hotspots. In PCCP random
view port for security enhancement is provided during
password creation, which forces users not to select any
Hotspots. . Random view ports guide the users to make things
difficult for attackers, who can use common guessable
selection within an image.
D. Database Anonymity
With the use of the homomorphism vulnerability
algorithm the confidentiality information in the database is
maintained by denying the access to the database by the
unauthorized or third parties. The Id assigned to each node not
known to other nodes in the network using this complex data
can be shared effectively
E. Outcome of the testing
Expected outcome is to detect with less Attack load for
detection and with more accuracy (less false positive and less
false negative) while making the detection coverage high
which makes the test effective.
VI. CONCLUSION & FUTUREWORK
This paper explains some of the potential dangers that
occur due to the presence of XSS attacks and the security
problems in the web application. Among several injection
attacks Cross site scripting has major impact on web
applications. To test those threats many technologies are
available which has some inefficiency. Therefore in this work
we presented an efficient methodology to identify the
vulnerabilities by monitoring the behavior of the model.
The combination of persuasive cued click points & password guessing resistant protocol provide better results in
authentication system. Here we provide only one image for
the authentication purpose which is easier for the user to
remember and also requires less time, but for the attackers it is
very difficult to see at click point area in the image. PGRP is
more restrictive against brute force and dictionary attacks.
PGRP is apparently more effective in preventing password
guessing attacks, it also provide most convenient login
experience. In future we can add some empty click-points to
confuse the attacker and also make double click on specific
click points we can further improve the security.
REFERENCES
[1] Della-Libera, G., et al, Security in a Web Services World A Proposed Architecture and Roadmap, IBM corp, Microsoft corp, 7 , apr2002,
URL: http://msdn.microsoft.com/en-us/library/ms977312.aspx.
[2] Holgersson, j., and E.soderstrom, Web Service Security-Vulnerabilities
and Threats within the context of WS-Security, SIIT 2005.
[3] Morais A, and E.Martins, Injeo de Ataques Basedos em Modelo para
Teste de Protocols de Segurana, Thesis (Master in computer Science),
Institute of Computing, UNICAMP, State University of campians,
Brazil, 15, May 2009.
[4] Cachin, C., and J. Camenisch, Malicious and Accidental-Fault
Tolerance in Internet Applications: Reference Model and Use Cases,
LAAS, MAFTIA, 2000.
[5] Ladan MI, Web services: Security Challenges, in Proceedings of the
World Congress on Internet Security, 2011, WorldCIS11, IEEE Press,
Londres, Reino, Unido, 21-23, Feb 2011.
[6] SoapUI, [software], Version 4.5. Eviware, The Web Services testing tool
Security Testing Tool, URL:http://www.soapui.org.
[7] Lawrence, K., C.Laler, A, Nadalin, R.Monzillo, and P.Hallam-Baker,
Web Services Security: SOAP Message Security 1.1 (WS-Security 2006),
OASIS, 2006.
[8] Lawrence, K., C, Kaler, A. Nadalin, R. Monzilo, and P. Hallam-Baker,
Web Services Security: Username Token profile 1.1, OASIS, 2006.
[9] Zhao G., W.Zheng, J.Zhao, and H.Chen, An Heuristic Method for
Web-Service Program Security Testing, In Proceedings of the 2009
Fourth China Grid Annual Conference, CHINAGRID 09, IEEE
Computer Society Press, Yantai China, Aug 2009.
[10] Cristian F., H. Aghili, R. Strong, and D. Volev, Atomic Broadcast:
From Simple Message Diffusion to Byzabtube Agreement. In
Proceedings of the Twenty-Fifth International Symposium on Fault-
Tolerant Computing, IEEE Computer Society Press, Pasadena-CA,
USA, June 1995.
[11] Myers G.J., C. Sandler, and T.Badgett, The Art of Software Testing, 3rd
ed., Wiley Publishing, New Jersey, USA, 2011.
[12] Valenti AW, and E. Martins, Testes de Robustez em Web Services
porMeio de Injeo de Falhas, Thesis, Institute of Computing,
UNICAMP, State University of Campinas, Brazil, jun 2011.
[13] Canfora G., and M. Penta, Service-Oriented Architectures Testing: A
Survey In software Engineering, Springer-Verlag, Berlin, Heidelberg,
2009.
[14] Zhou L, J. Ping, H.Xiao, Z. Wang, GeguangPu, and Z. Ding,
Automatically Testing Web Services Choreography with Assertions, In
Proceedings of the 12th international Conference on Formal
Engineering Methods and Software Engineering, ICFEM10, Springer-
Verlag, Berlin, Heidelberg,2010.
[15] Rogan D., OWASP WebScarabLite [Software], Version20070504-1631,
Open Web Application Security Project 2011, URL:
http://www.owasp.org/software/webscarab.html.
http://www.internationaljournalssrg.org/ncgctict-2015.htmlwww.internationaljournalssrg.orghttp://msdn.microsoft.com/en-us/library/ms977312.aspxhttp://www.soapui.org/
-
National Conference on Green Computing Trends in Information and Communication Technology (NCGCTICT-2015)
ISSN: 2349 - 641X www.internationaljournalssrg.org Page 7
A Novel Method For Face Recognition Using Gabor
Wavelet
R.Prema1, Dr. P.Shanmugapriya
2
1Assistant Professor and Research Scholar , Department of CSE , SCSVMV University, Kanchipuram.,
2Associate Professor and Head, Department of IT , SCSVMV University, Kanchipuram
Abstract - Gabor wavelets (GWs) are commonly used for
extracting features for various applications like object
detection, recognition and tracking. This paper
proposes and analyzes Gabor Wavelet and Eigenface
method for face recognition.
Index Terms Gabor wavelets , feature extraction, eigen face.
I. INTRODUCTION
Humans are very good at recognizing faces and complex patterns. Even a passage of time doesn't
effect this capability and therefore it would help if
computers become as robust as humans in face
recognition. Face recognition system can help in
many ways :
1) Checking for criminal records .
2) Enhancement of security by using
surveillance cameras in conjunction with
face recognition system.
3) Finding lost children's by using the images received from the cameras fitted at
some public places .
4) Knowing in advance if some VIP is
entering the hotel.
5) Detection of a criminal at public place.
6) Can be used in different areas of science
for comparing a entity with a set of entities.
7) Pattern Recognition.
Currently there are a several methods to achieve face
recognition. Among them we have the neural network
approach, the statistical approach - primarily based on
histograms, the multiresolutional approach, the
information theory approach, and the eigenface
approach.
We would be focusing on the Eigenface approach.
This method was originally suggested by Alex P.
Pentland and Matthew A. Turk from MIT in 1991.
This method consist on weighting the difference
between a given face image and a mean image, which
is obtained by averaging a predefined set of faces.
The training set is a group of face images from which
the mean face is calculated. Face recognition takes
place by linearly projecting the image to a low dimensional image space and weighting the
difference with respect to a set of eigenvectors. If the
difference (weight) is bellow certain threshold, the
image is recognized as a known face; otherwise, the
face can be classified as an unknown face, or not a
face at all.
Some of the limiting factors of this approach are the
background, difference in illumination, imaged head
size, and head orientation. To solve some of these
problems we could identify the location of the head
and zoom until we observe most of the face. We
could also set the camera's lighting based on the time
of the day.
In this paper, we propose a eigen face approach
with Gabor wavelet Transform (GWT) for feature extraction.
This paper is organized as follows: Section II
describes and Face Recognition system also describes
and analyzes eigenface approach and Gabor wavelet
Transform(GWT) . Section III shows results. The
concluding remarks will be given in Section IV.
http://www.internationaljournalssrg.org/ncgctict-2015.htmlwww.internationaljournalssrg.org
-
National Conference on Green Computing Trends in Information and Communication Technology (NCGCTICT-2015)
ISSN: 2349 - 641X www.internationaljournalssrg.org Page 8
II. FACE RECOGNITION SYSTEM
A facial recognition system is a computer application
for automatically identifying or verifying a person
from a digital image or a video frame from a video source. One of the ways to do this is by comparing
selected facial features from the image and a facial
database. Although humans perform face
recognition in an effortless manner, underlying
computations within the human visual system are
of tremendous complexity. The seemingly trivial
task of finding and recognizing faces is the result
of millions years of evolution and we are far away
from fully understanding how the brain performs this
task. Up to date, no complete solution has been
proposed that allow the automatic recognition of
faces in real images. In this section we will
review the face recognition systems using Gabor
Wavelet Transform (GWT).
Fig.1 Flow chart of Feature Extraction Stages
http://www.internationaljournalssrg.org/ncgctict-2015.htmlwww.internationaljournalssrg.orghttp://en.wikipedia.org/wiki/Application_softwarehttp://en.wikipedia.org/wiki/Identification_of_human_individualshttp://en.wikipedia.org/wiki/Authenticationhttp://en.wikipedia.org/wiki/Personhttp://en.wikipedia.org/wiki/Digital_imagehttp://en.wikipedia.org/wiki/Film_framehttp://en.wikipedia.org/wiki/Videohttp://en.wikipedia.org/wiki/Facehttp://en.wikipedia.org/wiki/Database_management_system
-
National Conference on Green Computing Trends in Information and Communication Technology (NCGCTICT-2015)
ISSN: 2349 - 641X www.internationaljournalssrg.org Page 9
A. Gabor Wavelet Transform
Gabor wavelets are widely used in
image analysis and computer vision [8] .The
Gabor wavelets transform provides an effective way to analyze images and has
been elaborated as a frame for
understanding the orientation and spatial
frequency selective properties of simple
cortical neurons. They seem to be a good
approximation to the sensitivity profiles of
neurons found in visual cortex of higher
vertebrates. The important advantages are
infinite smoothness and exponential decay
in frequency. Let be the gray level distribution of the input image, Gabor
wavelets transform on can be written as a convolution of with a family of kernels
k :
Where * denotes the convolution operator, and is the convolution result at k. The
Gabor wavelets (kernels) take the form of
a plane wave restricted by a Gaussian
envelope function [9]
vector enveloped by a Gaussian function,
where s is the standard deviation of this
Gaussian.
B. Feature vector generation
Feature vectors are generated at the
feature points as a composition of Gabor
wavelet transform coefficients. kth feature vector of ith reference face is defined as,
(3)
While there are 40 Gabor filters, feature
vectors have 42 components. The first two
components represent the location of that
feature point by storing (x, y) coordinates.
Since we have no other information about
the locations of the feature vectors, the first two components of feature vectors are very
important during matching (comparison)
process. The remaining 40 components are
the samples of the Gabor filter responses at
that point.
Although one may use some edge
information for feature point selection, here
it is important to construct feature vectors as
the coefficients of Gabor wavelet transform.
Feature vectors, as the samples of Gabor
wavelet transform at feature points, allow representing both the spatial frequency
structure and spatial relations of the local
image region around the corresponding
feature point.
By selecting different center
frequencies and orientations, we can generate a
family of GW kernels using (3), which is then used for feature extraction from images. Given
a gray-level image I (x, y), the GW features are
extracted by convolving I (x, y) with each of the GWs, as in (2).
Fig.2 Facial feature points found as the high-
energized points of Gabor wavelet responses
The Convolution can be computed efficiently using FFT, then point-by-point multiplica-tions, and
finally the inverse FFT (IFFT). By concatenating the convolution outputs, we can obtain a GW feature
http://www.internationaljournalssrg.org/ncgctict-2015.htmlwww.internationaljournalssrg.org
-
National Conference on Green Computing Trends in Information and Communication Technology (NCGCTICT-2015)
ISSN: 2349 - 641X www.internationaljournalssrg.org Page 10
vector.
C. Similarity Calculation
In order to measure the similarity of two
complex valued feature vectors, following similarity
function is used which ignores the phase:
vi, k (l) vt , j (l) (4)
feature vector of ith reference face, (vi,k), where l is
the number of vector elements.
Proposed similarity measure between two vectors
satisfies following constrains:
0 < Si< 1,
and if ith gallery face image is used also as the test
image,
Si ( j, j)< 1.
The location information is not used for vector
similarity calculation, but only the magnitudes of the
wavelet coefficients are take place at (3). It must be
clarified that the similarity function (3.8) is only one
component of the proposed matching procedure .
Location information of feature vectors will also be used during matching.
Equation (3) is a very common similarity
measure between feature vectors, containing Gabor
wavelet transform coefficients [36], but sometimes
we might have small variations [23, 27]. In [23]
similarity function at (3) is used with complex
valued coefficients and an additional phase
compensating term. In the early experiments it is
observed that small spatial displacements cause
change in complex valued coefficients due to phase rotation. Then phase can either be ignored or
compensated as in [23]. Although phase
compensated similarity function is found to increase
recognition performance significantly [23,27],
similarity function without phase is chosen to avoid
computational complexity.
D. Face comparison
After feature vectors are constructed from
the test image, they are compared to the feature
vectors of each reference image in the database. This
comparison stage takes place in two steps. In the
first step, we eliminate the feature vectors of the
reference images which are not close enough to the feature vectors of the test image in terms of location
and similarity. Only the feature vectors that fit the
following two criterions are examined in the next
step.
(5)
where th1 is the approximate radius of the area that
contains either eye, mouth or nose, (xr, yr) and (xt, yt)
represents the location of a feature point on a
reference face and test face respectively. Comparing
the distances between the coordinates of the feature
points simply avoids the matching of a feature point
located around the eye with a point of a reference facial image that is located around the mouth. After
such a localization, we may disregard the location
information in the second step. Moreover here
topology of face is also examined to use
corresponding information at the final matching by
only letting feature points that are match each other
in a topological manner.
Si(k,j)>th2, (6)
Similarity of two feature vectors is greater than th2,
where th2 is chosen as the standard deviation of
similarities of all feature vectors in the reference gallery and the similarity of two vectors is computed
by Equation (3).
III. RESULTS
http://www.internationaljournalssrg.org/ncgctict-2015.htmlwww.internationaljournalssrg.org
-
National Conference on Green Computing Trends in Information and Communication Technology (NCGCTICT-2015)
ISSN: 2349 - 641X www.internationaljournalssrg.org Page 11
In the following section, detailed information for
those four face databases and their corresponding
performance results for the proposed face
recognition method are given with the comparisons
with some major face recognition methods.
(a) (b)
Fig.3 Examples of different facial expressions of two people from Stirling database, a) gallery
faces, b) probe faces.
Table I: Recognition performances of eigenface, eigenhills and proposed method on the Purdue
face database.
Method Recognition rate (%)
Eigenface [20] 82.3
Eigenhills [31] 89.4
Proposed face recognition 100.0
method using GWT
http://www.internationaljournalssrg.org/ncgctict-2015.htmlwww.internationaljournalssrg.org
-
National Conference on Green Computing Trends in Information and Communication Technology (NCGCTICT-2015)
ISSN: 2349 - 641X www.internationaljournalssrg.org Page 12
IV.CONCLUSION AND FUTURE WORK
This method compares faces in terms of mouth,
nose and any other features rather than eyes. Moreover,
having a simple matching procedure and low computational cost proposed method is faster than
elastic graph matching methods. Proposed method is
also robust to illumination changes as a property of
Gabor wavelets, which is the main problem with the
eigen face approaches. There is no training as in many
supervised approaches, such as neural networks. A new
facial image can also be simply added by attaching new
feature vectors to reference gallery while such an
operation might be quite time consuming for systems
that need training.
Although recognition performance of the
proposed method is satisfactory by any means, it can
further be improved with some small modifications
and/or additional pre-processing of face images. Such
improvements can be summarized as;
Since feature points are found from the responses of image to Gabor filters separately, a set of weights
can be assigned to these feature points by counting
the total times of a feature point occurs at those
responses.
A motion estimation stage using feature points
followed by an affine transformation could be
applied to minimize rotation effects.
In this paper, a new approach to face
recognition with Gabor wavelets is presented. The
method uses Gabor wavelet transform for both finding
feature points and extracting feature vectors.
REFERENCES
[1] V. Bruce, Recognizing Faces. London: Erlbaum, 1988.
[2] G. Davies, H. Ellis, and E. J. Shepherd, Perceiving and
Remembering Faces, New York: Academic, 1981.
[3] H. Ellis, M. Jeeves, F. Newcombe, and A. Young, Aspects
of Face Processing. Dordrecht: Nijhoff, 1986.
[4] R. Baron, Mechanisms of human facial recognition, Int.
J. Man-Machine Studies, vol. 15, pp. 137-178, 1981.
[5] D. C. Hay and A. W. Young, The human face ,
Normality and Pathology in Cognitive function, A. W. Ellis
Ed. London: Academic, 1982, pp. 173-202.
[6] S. Carey, A case study: Face Recognition, Explorations in the Biological Language, E. Walker Ed. New York:
Bradford, 1987, pp. 175-201.
[7] S. Carey, R. Diamond, and B. Woods, The development of face recognition- A maturational component? Develop.
Psych., vol. 16, pp. 257-269, 1980.
[8] A. P. Ginsburg, Visual Information processing based on spatial filters constrained by biological data, AMRL tech.
Rep., pp. 78-129, 1978.
[9] A. G. Goldstein, Facial feature variation: Anthropometric Data II, Bull. Psychonomic Soc., vol.13, pp. 191-193,
1979.
[10] A. G. Goldstein, Face related variation of facial features:
Anthropometric Data I, Bull. Psychonomic Soc., vol.13,
pp. 187-190, 1979.
[11] L. D. Harmon, The recognition of faces. Scientific American, vol. 229, pp. 71-82, 1973.
[12] D. Perkins, A definition of caricature and recognition,
Studies in the Anthropology of Visual Commun., vol. 2, pp.
1-24, 1975.
[13] J. Sergent, Microgenesis of face perception, Aspects of Face Processing, H. D. Ellis, M. A. Jeeves, F. Newcombe,
and A. Young Eds. Dordrecht: Nijhoff, 1986.
[14] T. Kanade, Picture processing by computer complex and recognition of human faces. Technical report, Kyoto
University, Dept. of Information Science, 1973.
[15] S. Lin, S. Kung, and L. Lin, Face Recognition / Detection
by Probabilistic Decision-Based Neural Network, IEEE
Trans. Neural Networks, vol.8, pp.114-132, 1997.
[16] R. Brunelli, T. Poggio, Face Recognition: Features vs. Templates, IEEE Trans. on PAMI, Vol. 12, No. 1, Jan.
1990.
[17] M. H. Yang, N. Ahuja, and D. Kriegman, A survey on face detection methods, IEEE Trans. On Pattern analysis
and Machine Intelligance, to appear 2001.
[18] S. Ranganath and K. Arun, Face Recognition Using
Transform Features and Neural Network, Pattern
Recognition, vol. 30, pp. 1615-1622, 1997.
[19] S. Lawrence, C. Giles, A. Tsoi, and A. Back, Face Recognition: A Convolutional Neural Network Approach,
IEEE Trans. on Neural Networks, vol. 8, pp. 98-113, 1997.
[20] M. Turk and A. Pentland. Eigenfaces for recognition. Journal of Cognitive Science, pp.71-86, 1991.
[21] P. Belhumeur, J. Hespanha, and D. Kriegman, Eigenfaces vs. Fisherfaces: Recognition using class Specific linear
projection, IEEE Trans. on PAMI, vol.19, no.7, 1997.
[22] B. Moghaddam, C. Nastar, and A. Pentland, Bayesian
Face Recognition using Deformable Intensity Surfaces,
IEEE Conference on CVPR, San Francisco, CA, June
1996.
[23] L. Wiskott, J. M. Fellous, N. Kr ger and Christoph von der Malsburg,
Face Recognition by Elastic Graph Matching, In
Intelligent Biometric
http://www.internationaljournalssrg.org/ncgctict-2015.htmlwww.internationaljournalssrg.org
-
National Conference on Green Computing Trends in Information and Communication Technology
(NCGCTICT-2015)
ISSN: 2349 - 641X www.internationaljournalssrg.org Page 13
Mobile Cloud Computing for Face Recognition in Social
Media and Future for Mobile apps
P.Sriranjani, UG Scholar, Department of IT, Anna University, BIT Campus, Tiruchirappalli-620024,Tamil Nadu,India.
Abstract
In todays lifestyle social networking has become easily receive
information about everyone in the world. In day-to-day life
mobile apps and mobile devices are developing rapidly. In
Android mobile devices we can detect person identity using
cloud computing which utilized the Face.com API(application
programming interface).The paper presents the face
recognition,mobile application and approaches to overcome the
challenges.we also applied Augmented Reality as an information
viewer to user.The result of testing shows that the system is able
to recognize face samples with the average percentage of 85%
with the total computation time for the face recognition system
reached 7.45 seconds,and the average augmented reality
translation time is 1.03seconds to get someones information.
KEYWORDS
Cloud Computing, Social Network, Face Recognition, Face
Detection, Augmented Reality, Mobile Cloud Computing
Challenges.
1. INTRODUCTION
Cloud computing refers to the use of networked infrastructure software and capacity to provide resources to users in an on-demand environment. With cloud computing, information is stored in centralized servers and cached temporarily on clients that can include desktop computers, notebooks, handhelds and other devices. Cloud computing exists when tasks and data are kept on the Internet rather than on individual devices, providing on-demand access. Applications are run on a remote server and then sent to the user. Mobile cloud computing is the form of cloud computing in combination with mobile devices. Mobile devices are increasingly becoming an essential part of human life as the most effective and convenient communication tools which is not restricted by time and place. However, the mobile devices are facing many challenges in their resources (e.g., battery life, storage, and bandwidth) and communications (e.g., mobility and security). Currently, social networking has become a very popular media for
many people . A lot of personal information can be obtained from
this social network. Search for detailed-identity can be easily
conducted through searching using the search engine sites or
existing social networking website. However, this appears to be a
less effective and the search must be performed in front of a
computer or laptop. The development of mobile technologies such
as smartphone and tablet allows a person to easily run a variety of
multitasking activities including basic activities of the phones to
run various applications such as email, multimedia, office
T.Palaniyammal, UG Scholar, Department of IT, Anna University, BIT Campus, Tiruchirappalli-620024,Tamil Nadu,India.
applications, etc. In addition, there are a variety of services on the
internet which is integrate their services with social network
Services. This condition was due to the current trend in it which is
aimed the capability for sharing the core of social network. Today,
most services on the internet must provide facilities for users to
connect their services with a popular social network services,
especially Facebook and Twitter. Start from this case, we designed
a system that can connect a person with a variety of services
through someones identity. Our proposed system uses a persons
face as the primary identity. This is because in general, we know
someone from his face. Our system has made facial identification
to know the various
social network and other internet activities followed by that person.
By utilizing the camera system on mobile devices, system
identification can be done by a person using face recognition
system. One thing to be considered in the application of facial
recognition systems is face recognition system requires high
computation. In this case, the mobile devices has only limited
resources and there is a problem in implementing face recognition
application in mobile devices. The work of show the advantage of
the computing process of facial recognition systems which is done
outside the server (cloud computing). The implementation of cloud
computing technology on mobile devices is aimed at creating
effective computation on mobile devices for performing process of
face recognition. In this work, the problem of interest is the design
and implementation of face recognition module on mobile devices
associated with the process of cloud computing. The design and
implementation is done by making an application to perform face
detection system on Android mobile device (onloading) including
augmented reality module and performing face recognition module
using clouds services. 2. LITERATURE REVIEW
This section will describe the explanation of the face recognition
system, cloud computing technology, Face.com API, and
augmented reality concept.
2.1. Face Recognition
Face recognition system is a system that performs engineering
method in an image to search for the identity or the information
contained in an image. Facial recognition systems are generally
divided into two stages . The first stage is the face detection
module which is the early stage (pre-processing). Then, is followed
by the facial recognition stage. Several techniques that can be used
to detect face in an image are :
a. Knowledge-based method
b. Feature invariant approaches
c. Template matching methods
d. Appearance based methods
Meanwhile, the face recognition system is a system designed on a
computer. This system is created to help identifying a person's face
through the image or digital video. One of the method commonly
used in the face recognition system is a way to compare the facial
feature of an image with a database of faces that have been taken
earlier. This is shown in Figure 1, which describe the flow of the
process official recognition
http://www.internationaljournalssrg.org/ncgctict-2015.htmlwww.internationaljournalssrg.org
-
National Conference on Green Computing Trends in Information and Communication Technology
(NCGCTICT-2015)
ISSN: 2349 - 641X www.internationaljournalssrg.org Page 14
Figure 1 Face Recognition Workflow
2.2. Cloud Computing Technology
Cloud computing system is a combination of computer technology
as a processor utilization and development of internet-based
computing in which information technology-related capabilities
provided as a service or on demand . One of the cloud computing
service is Google App Engine (GAE). GAE is a Platform as a
Services in cloud computing for building and hosting our web-
based applications in Google data center . Applications will be
built on a multiplevirtualized servers. Google App Engine provides
an on-demand services, that the resource will be used according to
the need of users. GAE has an automatic system to adjust resource
for applications that are built to increase the existing demand.
2.3. Face.com API
Face.com is an Application Programming Interface (API) services
for face recognition process. Face.com can be classified into PaaS
cloud computing service . Face.com provides an API for
developers to develop software that implement a face recognition
function. The process of facial recognition with Face.com API as
well as face recognition process in general. First, the user must
perform training to the new faces. In this training process, there is
a face detection stage and the result will be saved. Once the face
has successfully trained, the face can be recognized by the system.
2.4. Augmented Reality
Augmented Reality has been there for a long time, but has been
established as a research area in 1990s. There are many
definitions of augmented reality, however the general assumption
is that the augmented reality enables an enriched perspective by
superimposing virtual objects on the real world in a way that
persuades the viewer that the virtual object is a part of the real.
Therefore, augmented reality is a crossover between the real and
virtual world. Generally, augmented reality systems are divided
into two types :
a. Augmented reality based on marker.
A method that utilizes an illustration of a black marker and a
square-shaped anthers with a thick black border and white
background. Through the position faced with a computer camera,
the computer will make the process of creating 2D or 3D virtual
world.
b. Augmented reality without marker (markerless AR).
AR method does not require a marker to show the elements of the
virtual world when combined with a real-world environment. The
use of the markerless method is commonly used for face tracking,
object tracking and 3D motion tracking.
Figure 2 Augmenting virtual object in the real world image
Figure 2 depicted the process of combining real world and virtual
object for marker based AR. Since the begining of Augmented
Reality (AR) systems, the potential of collaborative AR was
exploited for different activities such as in military, in football
match or in the helmet of the pilot. They can see several pieces of
information. Nowadays some elements of AR are used for mobile
phone applications . By just pointing the camera to an object, we
will immediately receive information about the object on the
screen. In this paper, we implement augmented reality without
marker on Android platform, but we use the face as a primary
marker for augmented reality display.
3. PROPOSEDMETHOD
The system in this paper was designed to combine the computation
which is run on mobile device and the advantage of cloud
computing as explained . The computation on mobile device
(onloading) will perform the face detection module and augmented
reality to interact with the user. The other computation will run in
the cloud server (offloading) using Google App Engine (GAE) dan
Face.com service. Actually, the system in this paper was design in
three main modules including face detection on Android mobile
device, face recognition on cloud server using GAE and Face.com
API, and augmented reality as a result of a face recognition module
on mobile device. The overall system design is shown in Figure 3.
Figure 3 Block Diagram of System
Firstly, the system will perform a face detection through mobile
device (onloading). The face detection itself, will use native API
from Android called Android Face Detector API that detect face
from a bitmap. Then, the application on mobile device will make
the process of video stream which is directed to the face object
based on the operations selected by the user. After face detected,
mobile device will crop the image only on the face aspect as
described in Figure 4. The algorithm for cropping the face image is
as follows : Get the mid points of all the faces in the image, the
confidence value should be higher than 0.4 ; Calculate a rectangle
around the face which is about the dimensions as shown in Figure
4, the distance between the eyes is A ; Then, crop the image from
the coordinates that have just been calculated. After that, the
results of the face detection will be processed for the offloading of
the face recognition process to the cloud server by REST
http://www.internationaljournalssrg.org/ncgctict-2015.htmlwww.internationaljournalssrg.org
-
National Conference on Green Computing Trends in Information and Communication Technology
(NCGCTICT-2015)
ISSN: 2349 - 641X www.internationaljournalssrg.org Page 15
communication in the data network. Figure 5 shows the sequence
diagram of this system.
Figure 4 Illustration of Cropping Face Image
Figure 5 Sequence Diagram of System
The next step for this system is to perform face recognition module
on the cloud server. Figure 6 shows the block diagram of the face
recognition module. There are two Python scripts in the face
recognition module. Main.py is a script which has a function as a
connector with Android devices and also for Blobstore services
caller. Face_client.py is a script which has a function as an API
caller to the Face.com server. When the cloud server got the image,
the face image will be proceed to know the identity of the face
image. After the cloud server recognized the face image, it will
return the result to the mobile device with json response which
include, personal identity of the person. Then, mobile device will
decode the json response and show the result as an augmented
reality.
4. RESULT AND ANALYSIS
There are three main modules which has been tested from this
work i.e face detection module, face recognition module, and
augmented reality module. The result and analysis will be
described in the following sub-sections.
4.1. Face Detection
Face detection module in the mobile device is the first stage in the
main functionality of this face recognition application. The process
of face detection module work as a process of onloading that run
individually on the mobile device. The technical testing done in
this module consists of testing the camera resolution variations.
Testing of the variation is done by varying resolutions camera
ranging from low to high. The results of these variations in the
resolution of the camera that would be material to know the time
required to perform the face detection process in an image. Tests
conducted in the face detection module will be performed 10 times
for each condition of the camera resolution. The result of this
testing will be shown in Figure 7 that also implemented augmented
reality concept.
Figure 7 The Results of Face Detection Using Augmented Reality
concept.
The test results of face detection system for mobile devices are
implemented in two type that used low class (Galaxy GT S5570)
with a 600 MHz CPU specs, 384 Mb RAM, and quality of 3.15
megapixel camera and middle class (W I8150 Galaxy) which has
the specification of 1.4 GHz CPU, 512 Mb RAM, and quality 5
megapixel camera. When the testing process, it was found that the
working time of a face detection system in mobile devices
experience the difference in each variation of the camera
resolution. This is because the amount of work is a growing field
that is marked with greater resolution. When using a resolution of
240x160 pixels, the system can recognize the presence or absence
of the face with an average time of 0.46 seconds. When the
resolution was increased to 320x240 pixels, face detection system
in mobile devices requires a longer working time. Greater use of
the resolution would affect the long absence of the processing time
of the face detection system. The processor speed also affect the
working of the face detection system on mobile devices. The
difference can clearly be seen in Figure 8.
Figure 8 Face Detection System With Variation of the Camera
Resolution
4.2. Face Recognition
The results of the face recognition that has been built will be
discussed in this chapter as shown in Figure 9. This test aimed at
obtaining an analysis of the level of success in the system to
perform its function. The number of face data used in this test is 50
faces, with details of three men and two women with 10 face each
data. Examples of faces used in this testing are presented in Table
1. In the tests performed with the variation of the number of
training data, we obtained the results as shown in Table 2. Table 1
Face Samples
Face Object Object 1 Object 2 Object 3 Object 4 Object 5
Face Image
http://www.internationaljournalssrg.org/ncgctict-2015.htmlwww.internationaljournalssrg.org
-
National Conference on Green Computing Trends in Information and Communication Technology
(NCGCTICT-2015)
ISSN: 2349 - 641X www.internationaljournalssrg.org Page 16
Figure 9 Result of Recognition Process on GAE
Figure 10 shows the effect of the number of training data on the
level of face recognition of our system. X-axis show the number of
training data, while the Y axis expresses the degree of face
recognition. With the number of training data is only one piece, the
recognition rate remains low at 20%. Along with the increasing
number of the training data, it increased the level of recognition.
When the training data of five pieces, the recognition rate reached
the average of 100% from 10 trials. Figure 11 shows the effect of
the number of data training to the result of error rates.
Table 2 Test Result with Variation of the Number of the Training
Data
In contrast to the recognition rate, error rate is obtained from the
variation of the number of training data which has a quite different
value. With only single training data, the error rate is obtained at a
value of 33%. The error values continue to decrease until the
training data is 4, with a value of 28% error. But when the number
of training data continues to be increased, the error rate to be
increase to 31.2%. After that the error will be returned to the
stability value of 30.2%. The differences in the recognition rate
and error rate, indicated that the number of training data affects the
value of recognition and error rate. When there are only few
training data, the learning system of the face will low. The more
training data, the better learning system.
Figure 10 The Effect of Total Training Data for Face Recognition
Effect
Figure 11 The Effect of Total Training Data for Face Recognition
Error Rate
4.3 Augmented Reality
This section we discuss the result of testing and analysis of our
system. The systems has been successfully developed on android
devices. The functionality of the whole system has been working
well. We tested to find out the weakness and error. Testing was
intended the system to find out the translation time of augmented
reality. Tests performed on normal lighting conditions using ten
samples and performed on the front face recognition process as
many as 10 attempts for each face with different expression. It is
intended to maximize the face recognition results obtained and to
get more detailed information on the face intended to be
recognized. The average time to test the result of the translation is
shown in Figure 12.
Figure 12 Average Translation Time
Based on the translation of the image obtained from 9 information
access, information that contains the complete data about the
owner of the familiar faces are much slower compared with an
access permission to display bio information. In displaying
detailed information the translation
time lasts for 1.03 seconds, while the bio-data to show the
translation time is 0.69 seconds. This happens because the system
http://www.internationaljournalssrg.org/ncgctict-2015.htmlwww.internationaljournalssrg.org
-
National Conference on Green Computing Trends in Information and Communication Technology
(NCGCTICT-2015)
ISSN: 2349 - 641X www.internationaljournalssrg.org Page 17
works by making the initial approach of face detection and face
recognition. Condition that occurs when the test was to ignore the
value of face recognition are issued either agree or disagree with
the input face image. Determination of the face affects the process
of facial recognition as the face detection results will serve as the
primary marker in our recognition. If there are objects that are
blocking the area of the face like a scarf or hair covering the
forehead, the face recognition algorithm will not deliver good
result as the translation was not the same for each time. The result
of translation that interact with social media is shown in Figure
Figure 13 Illustration of Social Media Interaction
In addition to the effect caused by the recognition process, it
should be noted that the information will be displayed in the
different request. Plain text files take longer to be parsed and
translated. JSON translation to spend a long time, to exchange data
with complete information requests because of size issue. JSON
object with complete information has a size greater than JSON
with a little information. The different is around a hundred bytes.
This impact on the variation of translation time.
4.4 Total Computation Time
This section explains the testing result of all modules that
implemented on mobile devices. This section discussed about the
computing process which was done on the face detection module
and the face recognition module. Computational processes that
occur in the face recognition systems shows the effectiveness of
the, especially when integrated with cloud computing technology.
The application of cloud computing technology to make the
process of offloading can be used to save energy. Here is an
overall result of the computation process of facial recognition
systems on mobile devices as shown in Table 3. Table 3
Description of Computational Time for Face Recognition System
on
MobileDevices
Figure 14 Total Computation Time for Different Device
Figure 14 shows that the computation to perform the face detection
process (onloading) require a faster time on all types of mobile
devices. This differs from the process of offloading that requires a
longer computation time. This is due to the process of offloading
the server cloud computing requires high computation and the
process of sending the result of face recognition which is also
highly dependent on network connectivity. However, the overall
process of face recognition is implemented on mobile device
which indicate the results are not so bad. The process of visual
tracking or face detection process that has been tested run well and
indicates the system works in real-time.
5.Advantages of Mobile Cloud Computing
and anytime.
location, context, and requested services to improve user
experience.
sing, and power
resources which are advantageous.
Computing such as solving the problem of WAN latencies by
using cloudlet.
Weiguang Song summarize the core concepts of Mobile Cloud
Computing [MCC] by developing a basic idea model of Mobile
Cloud Computing. Major problems faced by MCC are discussed
such as stability of wireless connectivity, tackling the unnecessary
battery usage etc. Also, few possible solutions are suggested.
Qureshi discusses about the mobile cloud computing technology
and proposes the implementation methods for Mobile Cloud
Computing solutions such as General Purpose Mobile Cloud
Computing (GPMCC) and Application Specific Mobile Cloud
Computing (ASMCC). Certain barriers such as network
availability and bandwidth are focused. Two aspects of security
issues such as mobile device security and cloud security are
addressed. Le Guan addresses the challenges in Mobile Cloud
Computing design such as network latency, limited bandwidth and
availability. In order to analyze Mobile Cloud Computing
technology, a concept model is proposed which includes context
management, resource scheduling, client and transmission channel.
A Cloud architecture of Mobile Cloud Computing is described for
organization of Mobile Cloud Computing systems. Application
partition and offloading and various context aware services are
explained briefly. Dejan addresses several mobile cloud
approaches. An overview of various possibilities of Mobile Cloud
Computing is given. Native and web applications are too extremes
of mobile applications. The cost model of elastic mobile cloud
applications is described. Han qi discuss Mobile cloud computing
(MCC) as a development and extension of mobile computing (MC)
and cloud computing (CC) which has inherited high mobility and
scalability. The proposed system in the paper explains the principle
of MCC, characteristics, recent research work, and future research
trends. Proposed system analyzes the features and infrastructure of
mobile cloud computing and also analyzes the challenges of
mobile cloud computing. Ashwin focuses on the capabilities of the
mobile and cloud landscape. New class of applications called
Cloud Mobile Hybrid [CMH] applications and a Domain Specific
Language [DSL] are defined. The proposed system define Cloud-
mobile hybrid as a collection of application that has a Cloud based
back-end and a mobile device front-end. Using a single DSL script,
proposed system is capable of generating a variety of CMH
applications. These applications are composed of multiple
combinations of native Cloud and mobile applications. The
proposed system also reduces the complexities of the platform.
Dejan discuss about the mobile communities which introduce new
requirements compared to traditional online web communities. On
the other hand, cloud computing is emerging as computing concept
that gives the computational resources on demand and abstraction
of technical details from the clients. The paper proposes Mobile
Community Cloud Platform (MCCP) as a cloud computing system
that can influence the full potential of mobile community growth.
http://www.internationaljournalssrg.org/ncgctict-2015.htmlwww.internationaljournalssrg.org
-
National Conference on Green Computing Trends in Information and Communication Technology
(NCGCTICT-2015)
ISSN: 2349 - 641X www.internationaljournalssrg.org Page 18
An analysis of the core requirements of common mobile
communities is provided. The paper presents the design of cloud
computing architecture that supports building and evolving of
mobile communities. Harshit presents a middleware for
distributing computation over mobile ad-hoc networks. Mobile
adhoc is used as an alternative for cloud in its absence. Synergy is
mainly used for energy conservation when the cloud is not
available, the battery life of mobile devices becomes dead hence
mobile ad-hoc is used as an alternative. The proposed system has
two applications such as prototype implementation of Synergy and
integrates OpenCV with it. Al though this is not stronger than
clouds, this must co-exist to improve the mobile computing
accessibility. Vinod discuss about the cloud computing which
enables the work anywhere anytime by allowing application
execution and data storage on remote servers. This is useful for
mobile computing and communication devices that are constrained
in terms of computation power and storage. The goal of the paper
is to characterize under what scenarios cloud-based applications
would be relatively more energy-efficient for users of mobile
devices. Hung analyzes the performance of many mobile
applications which are weak due to lack of computation resources,
storage, and bandwidth and battery capacity. To overcome this,
application is rebuilt using the cloud services. The proposed
system explains a framework to execute the mobile application in
cloud based virtualized environment with encryption, and isolation
to protect against unauthenticated cloud providers. Results show
the execution of mobile application by offloading the workload
with efficient application level migration method via mobile
networks. The migration of application form one device to another
is easy and quick in the proposed system. Ricky builds an elastic
mobile cloud computing infrastructure by introducing eXCloud
system. eXCloud is a middleware system which allows resources
to be integrated and used dynamically. In eXCloud, a Stack-on-
Demand (SOD) approach is used to support computation mobility
in the mobile cloud environment. The proposed system evaluation
shows that stack-on-demand model enhances state of the art by
increasing the computation and reducing migration overhead and
latency. Ricky discuss that mobile cloud computing allows mobile
applications to use the large resources in the clouds. In order to
utilize the resources, migration of the computation among mobile
nodes and cloud nodes is necessary. Therefore, a highly portable
and transparent migration approach is needed. The paper uses a
Java byte code transformation technique for task migration without
effecting normal execution. Asynchronous migration technique is
used to allow migrations to take place virtually anywhere in the
user codes. The proposed Twin Method Hierarchy minimizes the
overhead from state-restoration codes in normal execution. Milos
discusses the Biometric applications such as fingerprint
identification, face, or iris scanning. These applications actually
work in a laboratory setting where the client computer has
unlimited access to the throughput and computational resources of
the network. The problem focused here is on the battery power of
the device and the throughput of the communication channel of the
client node to the cloud. The paper explains the mobile cloud
computing technique for biometric applications such as fingerprint
identification, face recognition and iris recognition. Debessay
analyzes and studies the impact of cloudlets in interactive mobile
cloud applications. In order to study the impact, cloudlet network
and service architecture is proposed. This architecture focuses on
file editing, video streaming, and collaborative chatting. The
performance gains with the usage of clouds are shown by
simulation results. NKosi discusses mobile devices which are used
in Health information delivery access and communication
challenges like power, bandwidth, and security. The proposed
system explains how cloud computing can be used in mobile
devices to provide sensor signals processing and security. The
system described in the proposed system uses an NGN/IMS system
with cloud computing to reduce the burden of organizing and also
for improving the functions of existing mobile health monitoring
systems. The interaction between health service provider, IMS
network operator and cloud computing service providers should be
regulated so that identity management and security verification is
performed. Saeid describes the reviewed and synthesized
smartphone augmentation approaches. Generating high-end
hardware is more expensive, energy consuming and time-
consuming. Conserving local resources through Cyber Foraging
and Fidelity Adaptation are feasible and widely acceptable
approaches but they lack in providing data security. Reducing
resource requirements is achieved through cloud computing and
mashup technology. Peng propose a framework of Operational
Command Training Simulation System based on mobile cloud
computing. The system combines cloud computing and mobile
computing, which includes infrastructure, platform, support,
application and middleware layer. The detail design of middleware
layer has been explained in the paper. The problem of the mobile
terminal with limited resources has been solved, and the
distribution and interoperability of simulation systems were
enhanced. Yan Gu focuses on the fundamental issue in the mobile
application platform which is the deployment decision for
individual tasks when the battery life of the mobile device is a
major concern for the mobile users experience. The proposed
system explains the deployment scheme to offload expensive
computational tasks from thin, mobile devices to powered,
powerful devices on the cloud. The proposed system is
implemented and various experiments on the Android devices for
individual components. Chun discuss about the mobile applications
which are providing functionality on mobile devices. Also, mobile
devices provide strong connectivity with more powerful machines
ranging from laptops and desktops to commercial clouds. The
proposed system in the paper presents the design and
implementation of CloneCloud. CloneCloud is a system that
automatically transforms mobile applications to get benefit from
the cloud. CloneCloud uses a combination of static analysis and
dynamic profiling to automatically partition an application. Keerthi
discusses the services provided on the mobile devices which are
increasing day by day. One of the important services among them
is the Location Based Service (LBS). LBS depend on the
geographical position of the user to provide services to the end
users. A mobile device lacks in providing resources. Mobile device
should get resources from an external source, such as cloud
computing platforms. The main goal of the proposed system is to
provide dynamic location-based service. Srinivasa makes a
comparison on various existing web based operating systems. An
overview about proposed system is given along with the
architecture. Proposed platform is created by MeghaOS cloud
architecture and web browser which serves as both application
server and end user. MeghaOS offers services such as Account
manager, File manager, Message exchange etc. Many optimization
approaches are described. Chit propose a Mobile Computing
Applications Platform [MCAP] which is a cloud-enabled platform
for defining, developing, and deploying applications on smart
phones, tablets, and in-vehicle computers. Core services provide
support for location, user profile, notification, authentication,
content management, and device management. COTS technologies
for mobile computing and wireless networking are used to create a
low-cost and sustainable program. Yu-Jia proposes a secure frame-
work where the location information of mobile terminals is used in
a cloud computing environment. Various cloud capabilities have
made many application providers start migrating the data stored in
original databases to outsourced databases. The paper gives the
security model for location-based services and explains the use of
distributed storage and International Mobile Subscriber Identity
(IMSI) as user identification to secure the location data. An
enhanced privacy and authentication mechanism for the security
framework is also proposed.
6. CONCLUSIONS
In this work, we implemented a real-time image processing
application especially the face recognition system and user
interface on Android mobile device. We also created augmented
reality application on this project to provide information about the
recognizable faces viewed by the user. The result of testing
indicates that face recognition rate reached the average percentage
http://www.internationaljournalssrg.org/ncgctict-2015.htmlwww.internationaljournalssrg.org
-
National Conference on Green Computing Trends in Information and Communication Technology
(NCGCTICT-2015)
ISSN: 2349 - 641X www.internationaljournalssrg.org Page 19
of 85%. The use of face recognition system based on Augmented
Reality provides an attractive interface for the user. From the test
results, the translation time is 1.03 seconds to show the augmented
reality. This paper surveys the challenges, scope, approaches and solutions in the area of Mobile Cloud Computing. The paper focusses on Energy conservation in mobile devices, migration issues, application development platforms and the various mobile cloud computing applications. Face recognition system has been implemented using cloud computing technology (offloading
process) that uses REST to the cloud server communications which
are quite satisfactory with a fairly accurate result. However, the
overall system do not represent a real time system because 7
seconds is too long. For future development, the application of
cloud computing technology can be considered as an alternative to
save on computing in mobile devices along with its development is
quite extensive. So, the future, this system should be capable of
being a face recognition system in a real time as the development
of the communications network in providing a faster access.
REFERENCES
[1] Ashraf Jalal. The Use of Social Networking in Education :
Challenge and Opportunities. WCSIT.
2012.
[2] N. Balasubramanian, A Balasubramanian, and A.
Venkataramani, Energy Consumption in Mobile
Phone : a measurement study and implications for network
applications,ACM. 2009.
[3] A.P. Miettinen and J.K. Nurminen, Energy efficiency of
mobile clients in cloud computing,
HotCloud 2nd USENIX Workshop on Hot Topics in Cloud
Computing, 2010.
[4]
top related