DESIGN AND IMPLEMENTATION OF FACE DETECTION AND

RECOGNITION SYSTEM
PRELIMARY PAGES

Cover page

Title page

Approval page

Dedication

Acknowledgment

Table of content

Abstract

CHAPTER ONE

1.0 Introduction

1.1 Background of the research

1.2 Statement of research problem

1.3 Objectives of the study

1.4 Significance of the study

1.5 Scope of the study

1.6 Limitation of the study

1.7 Definition of terms

CHAPTER TWO: LITERATURE REVIEW

2.0 Introduction

2.1 Review of concept

2.2 Review of related work

2.3 Empirical Study

2.4 System Architectural Framework/Structure

2.5 Summary of the review

CHAPTER THREE: SYSTEM ANALYSIS AND DESIGN

3.1 Introduction

3.5 Problem of the existing system

3.2 Method of data collection

3.3 Data preparation

3.4 Program structure

3.6 Justification for the new system

3.7 System modeling

3.8 System flow chat

3.9 Activity diagram

3.10 Program flow chart

3.11 Database specification and design

CHAPTER FOUR: SYSTEM TESTING AND DOCUMENTATION

4.1 Introduction

4.2 Program language justification

4.3 Systems requirement

4.4 Implementation details

4.5 Procedure testing plan

CHAPTER FIVE

SUMMARY, CONCLUSION AND RECOMMENDATIONS

5.1 Summary

5.2 Conclusion

5.3 Recommendation

Bibliography

Source code
 CHAPTER ONE

1.0 INTRODUCTION

Face recognition system is an application for identifying someone from

image or videos. Face recognition is classified into three stages i.e) Face detection,

Feature Extraction, Face Recognition. Face detection method is a difficult task in

image analysis. Face detection is an application for detecting object, analyzing the

face, understanding the localization of the face and face recognition. It is used in

many application for new communication interface, security etc. Face Detection is

employed for detecting faces from image or from videos. The main goal of face

detection is to detect human faces from different images or videos. The face

detection algorithm converts the input images from a camera to binary pattern and

therefore the face location candidates using the AdaBoost Algorithm. The

proposed system explains regarding the face detection based system on AdaBoost

Algorithm . AdaBoost Algorithm selects the best set of Haar features and

implement in cascade to decrease the detection time .The proposed System for face

detection is intended by using Verilog and ModelSim,and also implemented in

FPGA.

Face Detection System is to detect the face from image or videos. To detect the

face from video or image is gigantic. In face recognition system the face detection

is the primary stage. Figure 1 shows the various stages of face recognition system
ie face detection, feature extraction and recognition. Now Face Detection is in vital

progress in the real world

Face recognition is a pattern recognition technique and one of the most important

biometrics; it is used in a broad spectrum of applications. The accuracy is not a

major problem that specifies the performance of automatic face recognition system

alone, the time factor is also considered a major factor in real time environments.

Recent architecture of the computer system can be employed to solve the time

problem, this architecture represented by multi-core CPUs and many-core GPUs

that provide the possibility to perform various tasks by parallel processing.

However, harnessing the current advancements in computer architecture is not

without difficulties. Motivated by such challenge, this research proposes a Face

Detection and Recognition System (FDRS). In doing so, this research work

provides the architectural design, detailed design, and four variant implementations

of the FDRS.

1.1 BACKGROUND OF THE RESEARCH

Face recognition has gained substantial attention over in past decades due to

its increasing demand in security applications like video surveillance and biometric

surveillance. Modern facilities like hospitals, airports, banks and many more

another organizations are being equipped with security systems including face
recognition capability. Despite of current success, there is still an ongoing

research in this field to make facial recognition system faster and accurate. The

accuracy of any face recognition system strongly depends on the face detection

system. The stronger the face detection system the better the recognition system

would be. A face detection system can successfully detect human face from a

given image containing face/faces and from live video involving human presence.

The main methods used in these days for face detection are feature based and

image based. Feature based method separates human features like skin color and

facial features whereas image based method used some face patterns and processed

training images to distinguish between face and non faces. Feature based method

has been chosen because it is faster than image based method and its’

implementation is far more simplified. Face detection from an image is achieved

through image processing. Locating the faces from images is not a trivial task;

because images not just contain human faces but also non-face objects in clutter

scenes. Moreover, there are other issues in face recognition like lighting

conditions, face orientations and skin colors. Due to these reasons, the accuracy of

any face recognition system cannot be 100%.

Face recognition is one of the most important biometrics methods. Despite the fact

that there are more reliable biometric recognition techniques such as fingerprint

and iris recognition, these techniques are intrusive and their success depends
highly on user cooperation. Therefore, face recognition seems to be the most

universal, non-intrusive, and accessible system. It is easy to use, can be used

efficiently for mass scanning, which is quite difficult, in case of other biometrics.

Also it is natural and socially accepted.

Moreover, technologies that require multiple individuals to use the same

equipment to capture their biological characteristics probably expose the user to

the transmission of germs and impurities from other users. However, face

recognition is completely non-intrusive and does not carry any such health

dangers.

Biometrics is a rapidly developing branch of information technology. Biometric

technologies are automated methods and means for identification based on

biological and behavioral characteristics of an individual. There are several

advantages of biometric technologies compared to traditional identification

methods. To take adequate measures against increasing security risks in modern

world, countries are considering these advantages and are shifting to new

generation identification systems based on biometric technologies.

1.2 STATEMENT OF RESEARCH PROBLEM

Biometric systems are becoming an important element (gateway) for

information security systems. Therefore biometric systems themselves have to

satisfy high security requirements. Unfortunately producers of biometric

technologies do not always consider security precautions. In publications regarding

biometric technologies, drawbacks and weaknesses of these technologies have

been discussed. Since biometrics form the technology basis for large scale and very

sensitive identification systems (e.g. passports, identification cards), the problem of

adequate evaluation of the security of biometric technologies is a current issue.

Also, some other issues with face detection and recognition system is on individual

with identical face like identical twins and others, in situation like this it is possible

for the system to make mistake or error in processing the person image so as to

grant access to the rightful user.

1.3 OBJECTIVES OF THE STUDY

The objective of this project is to implement a face recognition system

which first detects the faces present in either single image frames; and then

identifies the particular person by comparing the detected face with image database

or in the both image frames.

In addition to the main objective of this research work, the researcher also went far

more to add other features to the new system which are as fellow.

1. One of the objectives of this system is to design a system that will help the

organization maintain a strong security in the work environment.

2. Highlight areas of vulnerability in the new system

 3. Develop a ridged and secure database for the organization to enable them

secure their sensitive data and records.

1.4 SIGNIFICANCE OF THE STUDY

This study is primarily aimed at increasing efficiency in security, this research

work will help the users in maintaining data. This system will reduce the rate of

fraudulent activities as it can as well keep track of registered users and grant them

access upon face recognition completion.

Also the knowledge that would be obtained from this research will assist the

management to grow, also this research work will also be of help to the upcoming

researcher in this field of study both with the academic students on their study.

1.5 SCOPE OF THE STUDY

The scope of this study covers only on face detection and recognition,

accessing previous records and making matched for the data, updating of records

and making delete.

1.6 LIMITATION OF THE STUDY

 Many limitations encountered, were in the process of gathering information

for the development of this project work to this extent. It was not an easy one, so

many constraints were encountered during the collection of data.

The limitation focuses of the following constraints;

i. FINANCIAL CONTRAINTS: the cost of sourcing for information and data

that are involved in this work is high in the sense that we all know that information

is money.

ii. TIME: A lot of time was involved in writing and developing this work,

iii. Irregularities in power supply also dealt harshly with the researcher.

1.7 DEFINITION OF TERMS

Analysis: Breaking a problem into successively manageable parts for individual

study.

Attribute: A data item that characterize an object

Data flow: Movement of data in a system from a point of origin to specific

destination indicated by a line and arrow

Data Security: Protection of data from loss, disclosure, modification or destruction.

Design: Process of developing the technical and operational specification of a

candidate system for implements.

File: Collection of related records organized for a particular purpose also called

dataset.

Flow Chart: A graphical picture of the logical steps and sequence involved in a

procedure or a program.

Form: A physical carrier of data of information

Implementation: In system development-phase that focuses on user training, site

preparation and file conversion for installing a candidate system.

Maintenance: Restoring to its original condition

Normalization: A process of replacing a given file with its logical equivalent the

object is to derive simple files with no redundant elements.

Operation System: In database – machine based software that facilitates the

availability of information or reports through the DBMS.

Password: Identity authenticators a key that allow access to a program system a

procedure.

Record: A collection of aggregates or related items of a data treated as a unit.

Source Code: A procedure or format that allow enhancements on a software

package.

System: A regular or orderly arrangements of components or parts in a connected

and interrelated series or whole a group of components necessary to some

operation.
System Design: Detailed concentration on the technical and other specification that

will make the new system operational.

1.8 ORGANIZATION OF THE WORK

The project is organized in five chapters. With introduction already being

explained in chapter 1 and the whole idea of this research work presentation in

chapter one, like objective of the study, statement of the research area of coverage

limitation and definition of terms all this makes up the chapter one.

Chapter 2; this section deals with the review of study, review of concept theories

upon which this work is built on, the potential issues in the any face recognition

system in the form of difference in the lighting conditions in which the same

picture appears differently and the variations in skin color and pose.

Chapter 3 talks about the software tools used in the project mainly related to visual

basic programming language. The methodology at which this research work will

be implemented.

In chapter 4 the system is implemented and presented with its analysis. Functions

of the system and the operation of the system is also, in depth explained for reader

understating and comprehension. The system requirement is also detailed and the

platform at which the system can run on.

Chapter 5 summaries the whole work done and make possible recommendation

and suggest other points to be included into the work for future propose
REFERENCE

P. N. Belhumeur, J. P. Hespanha, and D. J. Kriegman, "Eigenfaces vs. Fisherfaces:

Recognition Using Class Specific Linear Projection", IEEE Transactions on

Pattern Analysis and Machine Intelligence, vol. 19, no. 7, 1997, pp. 711-720.

W. H. Press, S. A. Teukolsky and W. T. Vetterling, Numerical recipes in C: the art

of scientific computing, 2nd ed., Cambridge University Press, 1992. ISBN:0-521-

43108-5

I. Sajid, M.M. Ahmed, I. Taj, M. Humayun, F. Hameed, "Design of High

Performance FPGA based Face Recognition System", PIERS 2008 in Cambridge,

USA, 2-6 July, 2008.

R. Zhang, h. Chang, " A literature survey of face recognition and reconstruction

technique", Technical report, University of Texas, December, 2005.

M. Turk, A. Pentland, Eigenfaces for Recognition, Journal of Cognitive

Neuroscience, vol. 3, no. 1. 1991, pp. 71-86. [CrossRef]

Matsumoto T., Matsumoto H., Yamada K., Hoshino S., Impact of artificial

“gummy” fingers on fingerprint systems, in Optical Security and Counterfeit

Deterrence Techniques IV, vol. 4677 of Proceedings of SPIE, pp. 275–289, San

Jose, Calif, USA, January 2002.

 CHAPTER TWO

LITERATURE REVIEW

2.0 INTRODUCTION

Biometrics is a rapidly developing branch of information technology.

Biometric technologies are automated methods and means for identification based

on biological and behavioral characteristics of an individual.

This chapter focuses on the ongoing confronts in the field of the recognition and

some basic concepts of image applications, empirical study, review of related study

system architectural framework, challenges and the imaging concepts all will

described in detail un this chapter.

2.1 REVIEW OF CONCEPT

FACE IMAGE DETECTION MODEL

Face detection is the elementary step in the face recognition system and acts

as a stone to all facial analysis algorithms. Many algorithms exist to implement

face detection; each has its own weaknesses and strengths. The majority of these

algorithms suffer from the same difficulty; they are computationally expensive.

The image is a combination of color or light intensity values. Analyzing these

pixels for face detection is time consuming and hard to implement because of the

enormous diversity of shape and pigmentation in the human face. Viola and Jones
proposed an algorithm, called Haar-cascade Detector or called Viola-Jones, to

quickly detect any object, including human faces, using AdaBoost classifier

cascades that are based on Haar-like features and not pixels. Viola-Jones algorithm

is widely used in various studies involving face processing because of its real-time

capability, high accuracy, and availability as open-source software under the Open

Computer Vision Library (OpenCV) [8]. Viola-Jones detectors can be trained to

recognize any kind of a solid object, including human faces and facial features

such as eyes, and mouths. OpenCV has implemented Viola-Jones and provides a

pre-trained Haar-cascade for face detection.

INFORMATION MANAGEMENT

Data can be defined as individual facts or raw about something that can be

organized to generate useful information for decision-making. Information is

stimuli that have meaning in some contest for its receiver. When data is entered

into and stored in a computer, it IS generally referred to as information.

Graham (2001) said, with the move from local application to a web based ones,

also the data we created and access will need to undergo some profound changes.

Data and information undergoes a management process to maintain its consistence

and quality. Physical and logical security, quality assurance is emphasized to

ensure rational utilization and reliability of data information.

DATABASES

According to Lane and William (2004). a database is part of data

management system. They define a database as a container of data files, such as

product catalogs, inventories and item/customer records. They say that every

business would be a failure without a secure and reliable data management system.

They further say that information systems are the hearts of most businesses

worldwide, According to them. it is not easy to have a secure system, hut a system

developer must ensure that this is achieved. They advise system developers to have

clear subject areas, requirements and plans before they start designing the systems.

Braker and Hellerstein (1998), say that database has experienced a rapid increase

in growth since the development of a rational data base. The progress in database

systems and applications has produced a large scope of specialized technology

areas that have often become the exclusive domain of research specialists.

Examples include active database, object oriented, temporary database, and

deductive database areas. Advanced database systems were written by leading

specialists who have made significant contributions to the development of

technology areas.

Siau (2003) asserts that many databases that we find on the web today arc derived

from other databases. New databases arc often created because there is a need for
customized data and often, the databases are created with new data added in the

process. In this situation, a system that is able to carry along superimposed

information or annotation about the data is useful in many aspects. Siau (2003)

envisions an annotation management system that is not only capable of carrying

forward annotation of data being transformed. but that is also capable of attaching

new comments on derived data, back to the source data.

With French (1992), it’s also a single collection of structured data stored with a

minimum duplication of data item so as to provide a consistent and controlled pool

of data. This is common to all user systems, but is independent of the programs

that use data. The independence database and programs using it means that one can

be changed without changing the other. The user of database may find it

convenient to imagine that they are using integrated system.

According to Thierry (2006), the term database design can be used to describe

many different parts of design of an overall system Principally, and most correctly,

it can be thought as the logical design of the database of database structure used to

store data in are rational model these are the tables and views However the

database design could be sued to apply overall process of designing, and not just

base data structures, hut also forms and queries are used apart overall database

application within database management system (DBMS).

2.2 EMPIRICAL STUDY

.1 CHALLENGES IN FACE RECOGNITION

Over the years, face recognition has gained rapid success with the development of

new approaches and techniques. Due to this success, the rate of successful face

recognition has been increased to well above 90%. Despite of all this success, all

face recognition techniques usually suffer from common challenges of image

visibility. These challenges are lighting conditions variations, skin variations and

face angle variations. The challenges are explained in descriptive manner below.

.1.1 Difference in Lighting Conditions

The lighting conditions in which the pictures are taken are not always

similar because the variations in time and place. The example of lighting

variations could be the pictures taken inside the room and the pictures taken

outside. Due to these variations, a same person with similar facial expressions may

appear differently in different pictures. As a result, if the person has single image

store in the database of face recognition, matching could be difficult with the face

detection under different lighting conditions.

.1.2 Skin Color Variations

Another challenge for skin based face recognition system is the difference in the

skin color due to the difference in the races of the people. Due to this variation,

sometimes the true skin pixels are filtered out along with the noise present in an

image. In addition, the false skin background/non-background is not completely

filtered out during noise filtering process. So, it is a tough task is to choose the

filter which will cover the entire skin tones of different people and kick out false

skin noise.

.1.3 Variation in face angle or Orientation variation

The angle of the human face from camera can be different in different situations.

A frontal face detection algorithm can’t work on non-frontal faces present in the

image because the geometry of facial features in frontal view is always different

than the geometry of facial features in non-frontal view. This is why orientation

variation remains the difficult challenge in face detection system .

.2 IMPORTANT CONCEPTS OF IMAGING APPLICATIONS

.2.1 Image Processing

Image processing is a method of processing the image values, more precisely, the

pixels in case of digital images. The purpose of image processing is to modify the

input image such that the output image may change parametrically such as in

colors and representation. Image processing is the basic part of the face
recognition involving digital images. The processing can change the image

representation from one color space to another color space. It can also assign

different color values to targeted pixels for the purpose of keeping areas of interest

in output image. Image processing is also used to increase or decrease image

brightness, contrast and other morphological operations.

.2.2 Color Space

Color space is the representation of image colors in two or more color

components. Typical examples of color spaces are RGB, YCbCr and HSI color

spaces. In each of these color spaces the color of a pixel at any point in an image

is the combination of three color components. These color components vary from

0 to maximum value and this maximum value depends on the bits per pixel. The

different values in the range give different colors from black (0) to white (255)

with 8 bits per pixel.

.3.3 RGB color Space

RGB color space is the combination of red, green and blue color components. For

the 24 bits per pixel, the range of R, G and B varies from 0 to 255. If R, G and B

are all 0 then the resulted color will be black. If R, G and B are all 255 then the

output color will be white. The concept of the RGB color space is specified in

figure 2.b.3. Here the x-axis represents blue color range, y-axis represents green

color range and Z-axis represents red color range. As explained above, we can see
that black color is represented at the origin and white color is represented at the

other corner where red, green and blue are 255 each. Similarly we can have other

color values at different corners of cube corresponding to different RGB values.

Figure 2.2.3: RGB color model

2.3 REVIEW OF RELATED WORK

Guo proposed a two stage hybrid face detection system composed of the

probability based face mask pre-filtering and pixel based. Froba and Ernst[7]

proposed a face detector consist of 4 phase cascade structure based on MCT-

transformed images using the Adaboost learning algorithm. Knowledge-based

methods use facial features, such as two eyes, a nose and a mouth. Sung proposed

the feature invariant methods based on facial features such as invariant to pose,
lighting condition. The matching methods of the template are calculated by the

correlation between a test image and preselected facial templates.

Appearancebased, adopts machine learning techniques to extract features from a

pre-labeled training set. The Eigenface method is the most fundamental method for

finding the features.

McCready proposed a face detection method and implemented using nine

FPGA boards for the Transmogrifier-2 configurable hardware system. Sadri et al.

proposed neural network primarily based face detection on the Virtex-II Pro

FPGA. This face detection uses skin color filtering and edge detection to cut back

the processing time. Wei et al. [19] proposed, face detection using FPGA for

scaling input pictures and mounted-point expressions. The image size is simply too

small (120×120 pixels) solely some parts of classifier cascade are literally

implemented.

Wayman has studied the technical testing of biometric devices and divided it

into five subsystems: data collection, signal processing, transmission, data storage,

decision. This makes the potential attack points more clear. introduces three more

components: administrative management, information technologies, and

presentation of token. In total, 20 potential attack points and 22 vulnerabilities are

identified. All biometric systems require administrative supervision to some

extend. The level of supervision may vary for systems but it is not difficult to
imagine the related vulnerabilities. Vulnerabilities in this area may devalue even

the best planned system. A biometric system may or may not be related to an IT

environment, but is usually part of a larger system. Interaction with an IT

environment may introduce some new vulnerabilities not existing in the previous

scheme. A token is required in some biometric systems which make final decisions

based on presented biometric characteristic and information on the token. A token

may introduce a potential attack point to the biometric system. A smart card

containing biometric information is an example of a token used in this kind of

system. There are several other schemes for vulnerability classification [8].

Considering them as well, a generalized list of vulnerabilities of biometric systems

is suggested.

 Administration: Intentional or unintentional administrative mistakes.

 User: A legitimate user wants to upgrade his privileges to the administrative

level.

 Enrollment: Breaking registration procedures.

 Spoofing: A fake biometric is used for authentication as a legitimate user.

 Mimicry: Attacker mimics the biometric characteristics of the legitimate user.

 Undetect: Attacks undetected by the system may encourage new attacks.

 Fail secure: Result of abnormal utilization conditions of biometric system or IT

environment.
 Power: Power cuts.

 Bypass: Bypassing biometric system for access. This can be achieved by

surpassing physical barriers, forcing a legitimate user to present his biometric to

the sensor, or by cooperation of legitimate user.

 Corrupt attack – Weakening the system by making changes in the IT

environment or biometric system. Modification or replacement of system

parameters is an example.

 Degrade: Certain software in the IT environment decreases the system’s security

level.

 Tamper: Counterfeiting the hardware of the system.

 Residual: Latent fingerprints may be used to make artificial fingerprints or

accepted directly by the sensor.

 Cryptological attack: Encryption can be broken in data transmission and this

biometric data can be used for another type of attack (e.g. replay attack).

Consequently, there are many attack points and vulnerabilities in biometric

systems.

Using the given list of them, vulnerabilities for specific systems can be identified.

A biometric system may not have all of the vulnerabilities or attack points. The list

is general enough and can be applied to any system easily. For a specific system, it
is essential to consider the properties of the system in order to identify the

vulnerabilities.

The aim of vulnerability analysis is to determine the possibility of utilization of

weaknesses of biometric systems in an application environment. Penetration tests

are carried out to determine the vulnerability in the application environment of an

imposter with a certain potential of attack.

2.4 SYSTEM ARCHITECTURAL STRUCTURE/FRAMEWORK

This section describes the architectural design of FDRS, involve: the Mono

(sequential) and Parallel face recognition concepts.

The face recognition is the hardest algorithm because it has many steps before it

start the real recognition. A face must be detected to increase the possibility of

recognition and speed up the process by choosing one location in the image. To

detect a face, two steps must be done before the recognition. The first step is to

resize the image to standard size (determine by the administrator), apply some

filter to increase the quality, and convert the image into a compatible form. Next,

go to detection face, such that the image required to recognize is uploaded in the

memory with an Extensible Markup Language (XML) file to detect a face, and

finally, go to recognition step. In recognition, the extracted face will be compared

with training faces when they uploading to memory and extract face features by a

recognition algorithm.

Any operating system (OS) has multiple ways to deal with a process for different

structures. Some process has a single thread and other has multithreads architecture

(threads can run in a simultaneous manner).

Every biometric system is composed of four main modules:

Sensor module: A sensor acquires the biometric characteristic of an individual and

makes a digital description of it.

Feature extraction module: Input sample is processed and generates a

compressed image called template. Template is stored in database or in a smart

card.

Matching module: This module compares the presented biometric sample with the

template. In verification mode only one matching is performed resulting in only

one matching score and in identification mode the presented characteristic is

matched with many templates and generates many matching scores.

Decision module: This module accepts or rejects the user depending on the

matching score or security threshold.

Figure 1 presents such a system and possible attack points.

Figure 1: Attack points of a biometric system

Ratha et.al. identify eight attack points in this scheme. Let us shortly describe

characteristics of these attacks denoted by numbers 1-8 [5]:

1. Presenting a fake biometric sample to the sensor: A fake biometric sample such

as a fake finger, image of a signature, or a face mask is presented to the sensor in

order to get into the system.

2. Replay of stored digital biometric signals: A stored signal is replayed into the

system ignoring the sensor. For instance, replay of an old copy of a fingerprint

image or a recorded audio signal.

3. Denial of feature extraction: A feature set is formed by the imposter using a

Trojan horse attack.

4. Spoofing the biometric feature: Features extracted from input signal are replaced

by a fake set of features.

5. Attacking matching module: Attacks on matching module result in replacement

of matching scores by fake ones.

6. Spoofing templates in database: Database of saved templates can be local or

distant. The attacker tries to fake one or more biometric templates in the database.

As a result, either a fake identity is authorized or a rightful user faces a denial of

service.

7. Attacking the channel between the template database and matching module:

Stored templates are transmitted through a communication channel to the matching

module. Data in the channel can be changed by attacker.

8.Attacking the final decision process: If the final decision can be inserted or

blocked by the hacker then the authentication system function will be overridden.

Structure, architecture, production or implementation of a system may introduce a

vulnerability to the biometric system. In some cases a secondary system may be

integrated to the biometric system which possibly makes the biometric system

vulnerable. There are five points of vulnerabilities:

1. Operating systems;

2. Database management systems (and application software);

3. Biometric application software;

4. Software for sensor;

5. Hardware and drivers.

Other main aspects can be categorized as follows:

1. management of operations;

2. management of parameters (especially FAR/FRR parameters );

3. system configuration.

CPU Parallel Face Recognition

In the parallel face recognition process, two tasks can be done simultaneously. The

process of uploading training face images in the memory and the process of getting

face features from the training face images. The multithreading capability can be
Figure 3. CPU Parallel Face Recognition

2.5 SUMMARY OF THE REVIEW

FDRS has been implemented in four implementation variants (i.e., CPU

Mono, CPU Parallel, Hybrid Mono and Hybrid Parallel. Fisherface algorithm is

employed to implement recognition phase and Haar-cascade algorithm is employed

for the detection phase. In addition, these implementations are based on industrial
standard tools involve Open Computer Vision (OpenCV) version 2.4.8,

Microsoft .Net framework 4, VB programming language, EmguCV version

windows universal CUDA 2.9.0.1922, and heterogeneous processing units. The

experiment consists of applying 400 images for 40 persons' faces (10 images per

person), defining, training, and recognizing these images on these four variants, the

experiment is taken place on the same environment (laptop computer Intel core i7

processor 2.2 GHz, Nvidia GPU GeForce GT 630M, 7GB RAM). The speed up

factor is measured with respect to the CPU Mono implementation (the slowest than

all other three variants). The practical results demonstrated that, the Hybrid Parallel

Recognition is the fastest algorithm variant among the all, because it gives an

overall speed up around (82) times. The CPU Parallel gives an overall speed up

around (71). Finally, the Hybrid Mono gives a little improvement about (1.04).

Thus, employing parallel processing on modern computer architecture can

accelerate face recognition system.

REFERENCES

J. Ashbourn. Biometrics: Advanced Identity Verification. Springer-Verlag London,

UK, 2000.

W. Zhao, R. Chellappa, P. J. Phillips, and A. Rosenfeld, "Face Recognition: A

Literature Survey", ACM Computing Surveys, vol. 35, no. 4, 2003, pp. 399-458.

Full Text: Access at ACM

X. He, S. Yan, Y. Hu, P. Niyogi, and H. Zhang, "Face Recognition Using

Laplacian taces", IEEE Transactions on Pattern Analysis and Machine Intelligence,

vol. 27, no. 3, 2005, pp. 328-340.

P. N. Belhumeur, J. P. Hespanha, and D. J. Kriegman, "Eigenfaces vs. Fisherfaces:

Recognition Using Class Specific Linear Projection", IEEE Transactions on

Pattern Analysis and Machine Intelligence, vol. 19, no. 7, 1997, pp. 711-720.

A. Sharma, K.K. Paliwal, "Fast Principal component analysis using fixed-point

algorithm", Pattern Recognition Letters, Elsevier, vol. 28, no. 10, 2007, pp. 1151-

1155.

[CrossRef]
H. Anton, C. Rorres, Elementary Linear Algebra, ninth edition, Oxford University

Press, pp. 138.

R. L. Burden and J. D. Faires, Numerical analysis, seventh edition, Thomson,

ISBN:981-243-106-3.

L. Spacek, "Collection of Facial Images: Faces94",

http://cswww.essex,ac.uk/mv/allfaces/ (May 31, 2008).

Jain A. K., Uludag U., Attacks on biometric systems: a case study in fingerprints,

Proc. SPIE-EI 2004, Security, Seganography and Watermarking of Multimedia

Contents VI, San Jose, CA, pp. 622–633, 2004.

Jain A. K., Ross A., Uludag U., Biometric template security: challenges and

solutions, in Proceedings of the European Signal Processing Conference

(EUSIPCO '05), Antalya, Turkey,September 2005.

Roberts C., Biometric attack vectors and defenses, Computers and Security, vol.

26, no. 1, pp. 14–25, 2007.

Ratha N.K., Connell J.H., Bolle R.M., An Analysis of Minutiae Matching Strength,

Proc. 3rdAVBPA, Halmstad, Sweden, June 2001, pp. 223-228.

Wayman J.L., Technical Testing and Evaluation of Biometric Devices, in A. Jain,

et. al, Biometrics – Personal Identification in a Networked Society, Kluwer

Academic Publisher, 1999.

Cukic B., Bartlow N. The Vulnerabilities of Biometric Systems – An Integrated

Look at Old and

New Ideas, Technical Report, West Virginia University, 2005.

Dimitriadis C., Polemi D., Application of multi-criteria analysis for the creation of

a risk assessment knowledgebase for biometric systems. Lecture Notes in

Computer Science, Vol. 3072,

Springer-Verlag, ICBA, Hong Kong, China (2004), pp.724-730.

Lih-Heng C.; Sh-Hussain S. and Chee-Ming T. “Face Biometrics Based on

Principal Component Analysis and Linear Discriminant Analysis”, Journal of

Computer Science, Vol. 6, No. 7, pp. 693-699, 2010.

Wilson P. and Fernandez J. “Facial Feature Detection using Haar Classifiers”, The

Journal of Computing Sciences in Colleges, Vol. 21, No. 4, pp. 127-133, April

2006.

Runarsson K." A Face Recognition Plug-in for the PhotoCube Browser”, M.Sc.

thesis, Reykjavik University, December 2011.

Jafri R. and Arabnia, H. “A Survey of Face Recognition Techniques”, Journal of

Information Processing Systems, Vol. 5, No. 2, pp. 41-68, June 2009.

Xie, S. J.; Yang J.; Park, D. S. ; Yoon, S. and Shin, J. “State of the art in
biometrics” in Iris Biometric Cryptosystems, Yang, J. and Nanni, L., Eds.,
InTech, , Croatia, pp. 179-202, July 2011