Project Report On FAST MULTIMODAL BIOMETRIC APPROACH USING DYNAMIC FINGERPRINT AUTHENTICATION AND ENHANCED IRIS FEATURES

Submitted for partial fulfillment of the degree of B.E. (Computer Technology) By Anushree Sapre Apurva Jain Sanchita Bhriegu Shruti Sharma Under the Guidance of Ms. Ujwalla Gawande

Department of Computer Technology

Yeshwantrao Chavan College of Engineering, Nagpur

Rashtrasant Tukdoji Maharaj Nagpur University, Nagpur

2012-2013

YESHWANTRAO CHAVAN COLLEGE OF ENGINEERING NAGPUR (An Autonomous Institution Affiliated to RTMNU)

Department of Computer Technology

(2012-13)

Certificate
This is to certify that the Project Report titled Fast Multimodal Biometric Approach Using Dynamic Fingerprint Authentication and Enhanced Iris Features is submitted towards the partial fulfillment of requirement for the award of Degree of Bachelor of Engineering in Computer Technology awarded by Rashtrasant Tukdoji Maharaj Nagpur University, Nagpur.

Submitted by: Ms Anushree Spare Ms Apurve Jain Ms Sanchita Bhriegu Ms Shruti Sharma is approved. (Roll No: 102) (Roll No: 103) (Roll No: 111) (Roll No: 113)

Project Guide

Ms. Ujwalla Gawande

Project Coordinator Mrs. Gauri Chaudhary Head, Department of Computer Technology

Prof. A.R. Bhagat Patil

Date:___________

ii

Acknowledgements

Success is the manifestation of perseverance, inspiration and motivation. We the projectees, ascribe our success in this venture to all our lecturers without whom this project would have been a dream. We would like to thank Prof. Ujwalla Gawande who guided us throughout the project. It was because of her support and encouragement that this project materialized.

It was because of the timely guidance of Prof. A.R.Bhagat Patil that has helped us to complete the project against all odds. We are very graeful to him, for his inspiration, encouragement and guidance in all phases of our project. We also express our gratiude to all faculty members for their interminable support ans encougagement. Though words have their own limitaion we have made a modest effort to acknowledge the support extended.

We also take opportunity to thank our respected principal Dr. U.P. Waghe for his help in providing the necessary facilities for completion of project.

Finally, we would like to thank all the members of computer technology department for their co-operation ans support for timely completion of this project also we would like to thank our family and friends for their support and encouragement throughout the project.

iii

Abstract
Unimodal biometric systems have variety of problems such as noisy data, intra-class variations, restricted degree of freedom, non-universality, spoof attacks, and unacceptable error rates. Multimodal biometrics refers the combination of two or more biometric modalities in a single identification system. Biometric identification system based on the pattern of the human iris and fingerprint are well suited for a high level of security systems. However, the multimodal biometric system is limited to the time constraints due to its multiple processing stages. To overcome the problem of time taken we present a fast multimodal verification system by using the dynamic regions of the fingerprint image and enhanced iris segmentation method. This paper proposes enhanced fingerprint and iris recognition system that implements a fusion of both iris and fingerprint images at score level. The simulations are performed in the MATLAB environment to evaluate the performance of the implemented algorithms. Results and observations of the fusion are presented at the end.

iv

Table of Contents
1.0 Introduction.................................................1 1.1 Introduction 1.2 Finger Print Recognition 1.3 Iris Recognition 1.4Multimodal Biometric System

2.0 Theoretical Background / Literature survey....7

2.1 Literature Survey

3.0 Problem Definition.12

3.1 Problem Definition

4.0 Architecture and Design14

4.1 Architecture

5.0 Implementation Methodology..16

5.1 Database Used 5.2 Fingerprint Matching 5.2.1 Fingerprint Enhancement 5.2.1.1 Normalization 5.2.1.2 Segmentation 5.2.2 Minutiae Extraction 5.2.2.1 Minutia Marking 5.2.2.2 False Minutia Removal 5.3 Iris Recognition 5.3.1 Iris Segmentation 5.3.2 Iris Localization 5.3.3 Iris Normalization 5.3.4 Iris Feature extraction 5.4 Matching and Fusion

6.0 Experimental Results ..27

6.1 Experimental Results

7.0 Conclusion And Future Work33

7.1 Conclusion 7.2 Future Scope

Ennumerative Bibliography35

vi

List of Figures
Figure No.
1.1

Figure Name
Ridge ending, Bifurcation and short ridge

Page No. 4 4 5 15 19 20 21 21 21 21 23 24 24 25 28 29 29 30 30 31

1.2 1.3 4.1 5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9 5.10 6.1 6.2 6.3 6.4 6.5 6.6 6.7

Structure of Fingerprint Structure of an Iris

Schematic of score level fusion using hamming distance Normalization

Segmentation Termination Bifurcation Branch Minutiae Extraction False Minutia Removal Iris Segmentation Iris Localization Iris Normalization Snapshots of GUI Snapshots of GUI Snapshots of GUI Snapshots of GUI Snapshots of GUI Snapshots of GUI
rejection

Graphical representation of Image acceptance and 32

vii

List of Tables

Sr. No. 1

Table Accuracy Obtained 31

Page No.

viii

Our Publications
Title of the paper: Fingerprint-Iris Fusion Based Multimodal Biometric System Using Single Hamming Distance Matcher Authors: Prof. UjwallaGawande, AnushreeSapre, Apurva Jain, SanchitaBhriegu, Shruti Sharma Name of Journal:International Journal of Engineering Inventions (Online)

Other Details: e-ISSN: 2278-7461, p-ISSN: 2319-6491 (February2013) PP: 54-61

Volume 2, Issue 4

Submittted on:

10-feb-2013

Published on:

15-march-2013

ix

Chapter 1 Introduction

1.1 Introduction:
Biometrics are technologies used for measuring and analyzing a person's unique characteristics and is usually associated with the use of unique physiological characteristics to identify an individual. The paper is organized as follows. A typical biometric system is discussed ,with the need of multimodal biometrics is illustrated in Multimodal biometric system and different fusion techniques are mentioned with the description of score level fusion technique , challenges and conclusions are presented in the last section of the paper.

Biometrics is used for maintaining security at different levels. It is used in measuring features like face fingerprints, hand geometry, handwriting, iris, retinal, vein, voice etc. As the need of security is tremendously increasing and as well as transaction, frauds are increasing, the need for highly secure identification and personal verification technologies is becoming apparent. Prevailing methods of human identification based on credentials (identification documents and PIN) are not able to meet the growing demands for stringent security in applications such as national ID cards, border crossings, government benefits, and access control. As a result, biometric recognition, or simply biometrics, which is based on physiological and behavioral characteristics of a person, is being increasingly adopted and mapped to rapidly growing person identification applications. Although there are demerits in using biometric system such as noise in sensed data, intraclass variations, distinctiveness, non-universality, spoof attacks etc.

Above specified limitations imposed by unimodal biometric systems can be overcome by using multiple biometric modalities. Such systems, known as multibiometric systems, are expected to be more reliable due to the presence of multiple, fairly independent pieces of evidence. This approach also enables a user who does not possess a particular biometric identifier to still enroll and authenticate using other traits, thus eliminating the enrollment problems and making it universal. Further, if the biometric trait being sensed or measured is noisy (a fingerprint with a scar or a voice altered by a cold, for example), the resultant matching score computed by the matching module may not be reliable. This problem can be solved by installing multiple

sensors that capture different biometric traits. Such systems, known as multimodal biometric systems are expected to be more reliable due to the presence of multiple pieces of evidence. Multimodal systems also provide anti-spoofing measures by making it difficult for an intruder to spoof multiple biometric traits simultaneously. Multibiometric system performance is reliable because it has multiple information such as fusion, individual biometric feature extraction .The fusion of biometric is of 3 types i)Feature level fusion ii)Score level fusion iii)Decision level fusion. In our paper we have used score level fusion technique to combine the fingerprint and iris samples.

Challenges to Multi-Biometric System

Followings are the challenges in designing the multi modal system: 1. The information obtained from different biometric sources can be combined at different levels therefore selecting the best level of fusion will have the direct impact on performance and cost involved In developing a system. 2. There are Numbers of techniques available for fusion in multi-biometric system and the multiple source of information is available. Hence it is challenging to find the optimal solution for the application provided. 3. In multi-biometric systems the information acquired from different sources can be processed either in sequence or parallel. Hence it is challenging to decide about the processing architecture to be employed in designing the multi-biometric system.

1.2 Finger Print Recognition:

Fingerprint recognition or fingerprint authentication refers to the automated method of verifying a match between two human fingerprints. Fingerprints are one of many forms of biometrics used to identify individuals and verify their identity.The analysis of fingerprints for matching purposes generally requires the comparison of several features of the print pattern. These include patterns, which are aggregate characteristics of ridges, and minutia points, which are unique features found within the patterns.

The major Minutia features of fingerprint ridges are: ridge ending, bifurcation, and short ridge (or dot). The ridge ending is the point at which a ridge terminates. Bifurcations are points at which a

single ridge splits into two ridges. Minutiae and patterns are very important in the analysis of fingerprints since no two fingers have been shown to be identical.

Figure 1.1 Ridge ending, Bifurcation and short ridge(dot)

Figure 1.2 : Structure of fingerprint

One of the advantages of using fingerprint as one of the modalities is that fairly small storage space is requires for the biometric template, reducing the size of the database required. It is one of the most developed biometrics, with more history, research, and design. Each and every fingerprint including all the fingers are unique, even identical twins have different fingerprints. Sound potential for forensic use as most of the countries have existing fingerprint databases. Relatively inexpensive and offers high levels of accuracy.

1.3 Iris Recognition:

Iris Recognition is the process of recognizing a person by analyzing the random patterns of the iris. The automated method of iris recognition is relatively young existing in patent since 1994 only. The iris is the muscle in the eye that regulates the size of the pupil, controlling the amount of eye that enters the eye. It is the colored portion of the eye with the coloring amount based on the melatonin pigment present in the muscle. Although the coloration and structure of iris is genetically linked, the details of the pattern are not. The individual irises are unique and structurally different, which allows for it to be used for the recognition purposes.

Figure 1.3: Structure of Iris

1.4 Multimodal Biometric System:

Multimodal biometric systems are those that utilize more than one physiological or behavioral characteristic for enrollment, verification, or identification. In applications such as border entry/exit, access control, civil identification, and network security, multimodal biometric systems are looked to as a means of reducing false non-match and false match rates, providing a secondary means of enrollment, verification, and identification if

sufficient data cannot be acquired from a given biometric sample, and combating attempts to fool biometric systems through fraudulent data sources such as fake fingers.

The levels of fusion for multimodal systems are broadly categorized into three system architectures

Fusion at the Feature Extraction Level Fusion at the Matching Score Level Fusion at the Decision Level

In Fusion at the Feature Extraction Level, information extracted from the different sensors is encoded into a joint feature vector, which is then compared to an enrollment template (which itself is a joint feature vector stored in a database) and assigned a matching score as in a single biometric system.

In Fusion at the Matching Score Level, feature vectors are created independently for each sensor and are then compared to the enrollment templates which are stored separately for each biometric trait. Based on the proximity of feature vector and template, each subsystem computes its own matching score. These individual scores are finally combined into a total score which is passed to the decision module.

In Fusion at the Decision Level, a separate authentication decision is made for each biometric trait. These decisions are then combined into a final vote. This architecture is rather loosely coupled system architecture, with each subsystem performing like a single biometric system.

Chapter 2: Theoretical Background/Literature Survey

2.1 Literature Survey

A variety of articles can be found, which propose different approaches for unimodal and multimodal biometric systems. Many researchers have demonstrated that the fusion process is effective, because fused scores provide much better discrimination than individual scores. Such results have been achieved using a variety of fusion techniques.

Following papers have been taken for reference:

Mohamad Abdolahi, Majid Mohamadi, Mehdi Jafari (International Journal of Soft Computing and Engineering (IJSCE) ISSN: 2231-2307, Volume-2, Issue-6, January 2013) implemented a novel multimodal biometric system using fingerprint and iris with fuzzy logic. In this paper, they have saved fingerprint feature vectors in the form of 128 bit codes which consisted of 68 bit each of terminations and bifurcations. Instead of saving both the terminations and bifurcations, we are going to save the terminations(X and Y positions) in a 60 bit feature vector. The reason for not saving the bifurcation is that, in some of our images, there are no bifurcations at all or very few bifurcation points available. For extracting iris codes, in this paper, they have first obtained a rectangular region of the iris image with iris visible in the center. Then, the center pixel is obtained by dividing the row and column. Certainly the pixel is in the pupil region and its clear pupil is the darker part in eye, so it can move right to the pixel with a high amount of difference intensity and mark it, move left to the pixel with a high amount of difference intensity and mark it and find the center of these points.Same method is applied to find top and bottom and center of them. Now with these center and peripheral acquired points we can find the real pupil center with center point and maximum distance drawing a pupil circle performing the same task to find the iris region and extract iris from eye image. With Gabor filter features and iris code is extracted. We are going to use canny edge detection method for segmentation and Daugmains rubber sheet model for normalization. A fuzzy logic method is used for fusion which is given better performance and accuracy.Fuzzy logic is a kind of soft computing, which mimics human decision making. In our implementation, we are going to use a single hamming distance matcher to compute the final result.

A fast multimodal biometric system using fingerprint and iris fusion is proposed by A. Jameer Basha1, V. Palanisamy2, and T. Purusothaman3 (IEEE 2010). In this system, they have used reference point location and minutiae matching for fingerprint recognition and Daugmans rubber sheet model for iris recognition. For fusion, effective adaptive rank level fusion scheme that combine information presented by multiple domain experts based on the rank-level fusion integration method is employed. The ranks of individual matchers are combined using the highest rank approach.

Feten BESBES proposed a fusion based multimodal biometric system based on a couple of modalities recognition: fingerprint and iris, and every part provides its own decision. The final decision of the system will take in consideration both of the last decisions using the operator "AND".

Li Xiuyan1, Miao Changyun1, Liu Tiegen2, Yuan Chenhu3 (IEEE 2011) presented a theoretical analysis and experimental study on multimodal biometric. The theory and experiments of multimodal biometric were studied based on hand vein, iris and fingerprint. Simple Average and Weighting Average fusion algorithm, the classical information fusion methods, were analyzed and the constraint conditions for improving the recognition accuracy had been deduced. Biometric recognition experiments were performed finally to verify the theory deduction results. It is significant to future research on multimodal biometric and provides basis for developing multibiometric systems.

George Chellin Chandran, Dr. Rajesh. R.S (IJCSNS 2009) proposed Performance Analysis of Multimodal Biometric System Authentication that adopted multiple biometric traits of an individual, to establish the identity. The system employed multiple sensors to acquire data pertaining to fingerprint and iris. The independence of the traits ensures the improvement in performance. The main purpose of the proposed system was to reduce the error rate as low as possible and improve the performance of the system by achieving good acceptable rate during identification and authentication.

Adem Alpaslan ALTUN proposed Recognition of Selected Fingerprints and Iris FeaturesEnhanced by Curvelet Transform with Artificial Neural Networks In this study, curvelet transform is applied biometric images for enhancement. Obtained results after

applied curvelet transform is compared to the other traditional image enhancement algorithms. Features obtained from enhanced fingerprints and iris images are selected by using Genetic Algorithms because of too huge dataset. Selected features are input to Artificial Neural Networks for biometric recognition. Thus, the recognition is achieved very fast without to reduce the performance.

Teddy Ko (IEEE Computer Society 2005) researched on multimodal biometrics and came up with Multimodal Biometric Identification for Large User Population Using Fingerprint, Face and Iris Recognition that discusses the various scenarios that are possible in multimodal biometric systems using fingerprint, face and iris recognition. Here they have applied various levels of fusion techniques in order to generate output. KAZI M.M, RODE Y.S (Advances in Computational Research ISSN: 0975-3273 & EISSN: 0975-9085, Volume 4, Issue 1, 2012) proposed a multimodal biometric system using face and signature in which they used score level fusion. In this paper, matcher scores are simply added, with no prior normalization. Scores are neither rescaled, nor weighted to account for differences in matcher accuracy. Whereas in our implementation of fusion technique, first the individual score is calculated and the average of the two is considered as final score. If the score obtained is greater than a predefined threshold, then the output is accepted otherwise it is rejected.

Fingerprint Recognitionusing Image Segmentation proposed by SangramBana, Dr. DavinderKaur((IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES Vol No. 5, Issue No. 1, 012 023) is a study and implementation of a fingerprint recognition system based on Minutiae based matching quite frequently used in various fingerprint algorithms and techniques. Their approach mainly involves extraction of minutiae points from the sample fingerprint images and then performing fingerprint matching based on the number of minutiae pairings among two fingerprints in question. Here, they used a matching algorithm that was not upto the mark and gave reasonable results. One of the reasons for poor verification result was the bad quality of fingerprint images and inefficient matching algorithm as it is vulnerable to effects like scaling and elastic deformations.

10

To overcome the problems faced by individual traits, a novel combination is proposed for the recognition system. Our system improves the accuracy. The main drawback of multimodal system is that they are time comsuming our system aims at reducing the processing and making system fast using enhanced iris featurs and dynamic regions of fingerprint. The system has become fast because we have used single hamming distance matcher for both fingerprint features and iris features.

11

Chapter 3: Problem Definition

12

3.1 Problem Definition

The proposed project on multimodal biometric systems with fingerprint and iris recognition seeks to alleviate some of the problems of unimodal biometrics by providing multiple pieces of evidence of the same identity.

Conventional multimodal biometric identification systems tend to have larger memory footprint, slower processing speeds and a higher implementation and operational cost This project discusses the various scenarios that are possible to provide smaller memory footprint and to the improve the performance of multimodal biometric systems using the combined characteristics iris and fingerprint, the level of fusion (multimodal fusion) is applied to that are possible and the integration strategies that can be adopted in order to increase the overall system performance. In order to ensure that the performance of multibiometric systems such as fingerprint and iris will be powerful with respect to the quality of obtained fingerprint and iris images, these images are denoised and enhanced. Finally, the extracted feature vectors of both fingerprint and iris are fused together and Hamming distance is applied to identify the subject.

13

Chapter 4: Architecture and Design

14

4.1 Architecture

Figure 4.1 schematic of score level fusion using hamming distance .

15

Chapter 5: Implementation Methodology

16

5.1 Database Used

In order to check the validity of our implementation, we have used the database provided to us by our guide Prof. Ujwalla Gawande. Image Database consists of 50 subjects with four samples of fingerprint and iris each. Good quality images were taken.

5.2 Fingerprint Matching

There are two major for steps performed in order to match two fingerprint templates. These are as follows:

5.2.1 Fingerprint Enhancement

The performance of minutiae extraction algorithms and other fingerprint recognition techniques relies heavily on the quality of the input fingerprint images. In an ideal fingerprint image, ridges and valleys alternate and flow in a locally constant direction. However the fingerprint images obtained are usually poor due to elements that corrode the clarity of the ridge elements. This leads to problems in minutiae extraction. Thus, image enhancement techniques are necessary to reduce the noise and enhance the definition of ridges against valleys. In order to ensure good performance of the ridge and minutiae extraction algorithms in poor quality fingerprint images, an enhancement algorithm to improve the clarity of the ridge structure is necessary. The pre-processing steps include:

5.2.1.1 Normalization:

Normalization is done so that the gray level values lies within a given set of values. The fingerprint image is normalized to have a predefined mean and variance. This is required as the image usually has distorted levels of gray values among the ridges and the valleys. Normalization allows standardizing the distorted levels of variation in the gray scale values. Normalization involves pixel-wise operations and does not change the ridge and valley structures.

17

Normalization is a linear process. Suppose the intensity range of the image is 50 to 180 and the desired range is 0 to 255 the process entails subtracting 50 from each of pixel intensity, making the range 0 to 130. Each pixel intensity is multiplied by 255/130, making the range 0 to 255. The normalized image is given by N (i, j) = M0 + V0 (I (i, j) - M) 2/V if I (i, j) > M M0 - V0 (I (i, j) - M) 2/V otherwise

Where for a pixel I (i, j) the estimated mean and variances are M and V respectively. M0 and V0 denote the desired mean and variance values.

Histogram equalization, as normalization method, is a process to enhance the contrast of images by transforming its intensity values. Usually a fingerprint image has different gray values for every pixel. It is desirable to have the gray value around a mean value. This is achieved by histogram equalization. It increases the local contrast of images. Thus the intensities can be distributed on the histogram. This allows for areas of lower local contrast to gain a higher contrast without affecting the global contrast. Histogram equalization accomplishes this by effectively spreading out the intensity values.

The histogram of the original image illustrates that all the intensity values lie on the right hand side of the 0255 scale, with no pixels in the left hand side. The histogram of the normalized image shows that the range of intensity values has been adjusted such that there is a more balanced distribution between the dark and light pixels. Normalizing the image improves the contrast between the ridges and valleys. It does not alter the shape of the original histogram plot. The relative position of the values along the x axis is shifted.

18

Figure5.1 Normalization

5.2.1.2 Segmentation

In general, only a Region of Interest (ROI) is useful to be recognized for each fingerprint image. The image area without effective ridges and furrows is first discarded since it only holds background information. Then the bound of the remaining effective area is sketched out since the minutia in the bound region are confusing with those spurious minutia that are generated when the ridges are out of the sensor. Estimate the block direction for each block of the fingerprint image with WxW in size(W is 16 pixels by default). The algorithm is:

i)

Calculate the gradient values along x-direction (gx) and y-direction (gy) for each pixel of the block. Two Sobel filters are used to fulfill the task. For each block, use Following formula to get the Least Square approximation of the block direction. tg2 = 2 (gx*gy)/(gx2-gy2) for all the pixels in each block. The formula is easyto understand by regarding gradient values along x-direction and y-direction as cosine value and sine value. So the tangent value of the block direction is estimated nearly the same as the way illustrated by the following formula. tg2= 2sin cos /(cos2 -sin2 )

ii)

iii)

After finished with the estimation of each block direction, those blocks without significant information on ridges and furrows are discarded based on the following formulas:

E={2(gx*gy)+(gx2gy2)}/W*W*(gx2+gy2)

19

For each block, if its certainty level E is below a threshold, then the block is regarded as a background block.

Figure 2.2 : Segmentation

5.2.2 Minutiae Extraction

First step in minutiae marking is fingerprint thinning abd binarization. Thinning is done to eliminate the redundant pixels till the ridge is just one-pixel wide. A builtin MATLAB morphological thinning function is used here. Binarization is converting the image in the form of 0s and 1s.

5.2.2.1 Minutia Marking

After the fingerprint ridge thinning, marking minutia points is relatively easy. But it is still not a trivial task as most literatures declared because at least one special case evokes my caution during the minutia marking stage. In general, for each 3x3 window, if the central pixel is 1 and has exactly 3 one-value neighbors, then the central pixel is a ridge branch.If the central pixel is 1 and has only 1 one-value neighbor, then the central pixel is a ridge ending.

20

Figure 5.3: Bifurcation

Figure 5.4: Termination

Figure 5.5: Branch

Figure 5.5 illustrates a special case that a genuine branch is triple counted. Suppose both the uppermost pixel with value 1 and the rightmost pixel with value 1 have another neighbor outside the 3x3 window, so the two pixels will be marked as branches too but actually only one branch is located in the small region. So a check routine requiring that none of the neighbors of a branch are branches is added.

Where Pi is the pixel value in the neighborhood of P

Figure 5.6: Minutia Extraction

21

5.2.2.2 False Minutiae Removal

The preprocessing stage does not totally heal the fingerprint image. For example, false ridge breaks due to insufficient amount of ink and ridge cross-connections due to over inking are not totally eliminated. Actually all the earlier stages themselves occasionally introduce some artifacts which later lead to spurious minutia. These false minutia will significantly affect the accuracy of matching if they are simply regarded as genuine minutia. So some mechanisms of removing false minutia are essential to keep the fingerprint verification system effective. Following steps are used in order to remove false minutia:

i) If the distance between one bifurcation and one termination is less than D and the two minutiae are in the same ridge. Remove both of them. Where D is the average inter-ridge width representing the average distance between two parallel neighboring ridges.

ii) If the distance between two bifurcations is less than D and they are in the same ridge, remove the two bifurcations.

iii) If two terminations are within a distance D and their directions are coincident with a small angle variation. And they suffice the condition that no any other termination is located between the two terminations. Then the two terminations are regarded as false minutia derived from a broken ridge and are removed.

iv)

If two terminations are located in a short ridge with length less than D, remove the two terminations.

22

Figure5.7: False Minutiae Removal

5.3 Iris Recognition

Following steps are applied in order to extract iris feature vectors for matching:

5.3.1 Iris Segmentation The first stage of iris recognition is to isolate the actual iris region in a digital eye image. The iris region, can be approximated by two circles, one for the iris/sclera boundary and another, interior to the first, for the iris/pupil boundary. The eyelids and eyelashes normally occlude the upper and lower parts of the iris region. Also, specular reflections can occur within the iris region corrupting the iris pattern. A technique is required to isolate and exclude these artefacts as well as locating the circular iris region.

It was decided to use circular Hough transform for detecting the iris and pupil boundaries. This involves first employing Canny edge detection to generate an edge map. Gradients were biased in the vertical direction for the outer iris/sclera boundary. Eyelids were isolated by first fitting a line to the upper and lower eyelid using the linear Hough transform. A second horizontal line is then drawn, which intersects with the first line at the iris edge that is closest to the pupil.The second horizontal line allows maximum isolation of eyelid regions. Canny edge detection is used to create an edge map, and only horizontal gradient information is taken. The linear Hough transform is implemented

using the MATLAB Radon transform, which is a form of the Hough transform. If the

23

maximum in Hough space is lower than a set threshold, then no line is fitted, since this corresponds to non-occluding eyelids. Also, the lines are restricted to lie exterior to the pupil region, and interior to the iris region. A linear Hough transform has the advantage over its parabolic version, in that there are less parameters to deduce, making the process less computationally demanding.

Figure 5.8: Iris Segmentation

5.3.2 Iris Localization

Detecting and removing the occluding eyelashes is done by iris localization using circular hough transform algorithm.

Figure 5.9: Iris Localization

5.3.3 Iris Normalization

After successfully extracting the iris part from the eye image, in order to allow comparisons between different irises, transform the extracted iris region so that it has a fixed dimension, and hence removing the dimensional inconsistencies between eye images due to the stretching of the iris caused by the pupil dilation from varying levels of

24

illumination. Therefore, this normalization process will produce irises with same fixed dimensions so that two photographs for the same iris under different lighting conditions will have the same characteristic features. For normalisation of iris regions a technique based on Daugmans rubber sheet model was employed. The centre of the pupil was considered as the reference point, and radial vectors pass through the iris region. A number of data points are selected along each radial line and this is defined as the radial resolution. The number of radial lines going around the iris region is defined as the angular resolution. Since the pupil can be nonconcentric to the iris, a remapping formula is needed to rescale points depending on the angle around the circle.

Figure 5.10: Iris Normalization

5.3.4 Iris Feature Extraction This is the most key component of an iris recognition system and determines the systems performance to a large extent. Iris recognition produces the correct result by extracting features of the input images and matching these features with known patterns in the feature database. Features are the attributes or values extracted to get the unique characteristics from the image. Features from the iris image are extracted using Haar Wavelet decomposition process. In the wavelet decomposition the image is decomposed into four coefficient i.e., horizontal, diagonal, vertical and approximation. The approximation coefficients are further decomposed into four coefficients. The sequences of steps are repeated for five levels and the last level coefficients are combined to form a vector. The combined vector is binarized to allow easy comparisons between the iris codes for database and query image.

25

The binarized feature vectors are passed to the matching module to allow comparisons.

5.4 Matching and Fusion

The comparison is done between iris codes and fingerprint codes generated for database and query images using hamming distance approach. In this approach the difference between the bits of two codes of both are counted and the number is divided by the total number of comparisons. Where A is the binary vector for database image and B is the binary vector for query image while N is the number of elements. This matching score (MS) is used as input for the fusion module where the final matching score is generated.

No individual trait can provide 100% accuracy. Thus to overcome the problems faced by individual traits, a novel combination is proposed for the recognition system. The integrated system also provide anti spoofing measures by making it difficult for an intruder to spoof multiple biometric traits simultaneously. Scores generated from individual traits are combined at matching score level by calculating the average of the score obtained from individual modalities.

A predefined threshold is set based on the overall scores obtained. If the combined and averaged score is greater than the threshold then the person is genuine otherwise the person is considered as an imposter. In our implementation, we set the threshold to 70%.

26

Chapter 6: Experimental Results

27

6.1 Experimental Results

Final results are obtained by applying single hamming distance matcher individually for each feature vector, and then the scores obtained are fused by averaging features of both fingerprint and iris.

Figure 6.1 : Selection of fingerprint query image of subject 7

28

Figure 6.2: Accurate matching of fingerprint and iris image.

Figure 6.3: Result showing Person is genuine

29

Figure 6.4: Selection of fingerprint query image of subject 18

Figure 6.5: Inaccurate matching of fingerprint and iris image.

30

Figure 6.6: Result showing person is imposter due to incorrect matching

False Rejection Rate (FRR): For an image database, each query image is matched against the database images of the same finger to compute the False Rejection Rate. Same procedure is repeated for iris database.

False Acceptance Rate (FAR): Also the query image of each finger in the database is matched against the first sample of the remaining fingers to compute the False Acceptance Rate. Same procedure is repeated for iris database. In our implementation, the FAR and FRR was 35-40% approximately and the overall accuracy in the verification stage was about 70-75%.

Fingerprint Correct Match Incorrect Match Accuracy Rate 37 13 0.74

Iris 39 11 0.78

Fingerprint+Iris 37 13 0.74

Table 1: Accuracy Obtained

31

The graph below shows the acceptance and rejection of images based on their scores. On the X-axis are the images and on the Y-axis are the final matching scores.

Image Vs. Matching Score

80 70 60 50 40 30 20 10 0 1 3 5 7 9 1113151719212325272931333537394143454749 Series1

Figure 6.7: Graphical representation of Image acceptance and rejection

We also measured the time elapsed when using individual matchers and the fusion of the scores obtained using the MATLAB function tic; any statements; toc; the results obtained are as followsFingerprint matcher - elapsed time is 0.77715 seconds Iris matcher Fusion - elapsed time is 0.08923 seconds - elapsed time is 0.026365 seconds

Thus we see that fusion takes minimum time to calculate the final output.

32

Chapter 7: Conclusion and Future Scope

33

7.1 Conclusion
Thus, our multimodal system that uses dynamic fingerprint authentication and enhanced iris features provides faster and improved results because of the use of single hamming distance matcher used in the matching stage at the time of fusion. We observed that in order to extract correct feature vectors to be used for matching, the preprocessing steps must be applied accurately so that they remove the noise present in the image obtained at the image acquisition phase. Also, we felt that although hamming distance increases the computation speedup of our system, it is not the most accurate measure for matching. Various other distance matchers can also be used instead.

7.2 Future Scope

While our implementation is successfully able to decide whether the person is genuine or an imposter, it is by no means perfect. As shown by the results, there are erroneous results produced sometimes.

In order to make our implementation more efficient, there is scope in the matching and fusion module of our system. After extracting the correct feature vectors from both the modalities, it is essential to use correct matching and fusion algorithm to decide the genuineness of a person. During the implementation period we, felt that instead of hamming distance we could have obtained better results if we would have used other distance matchers for example Euclidean distance.

Another improvement would be to encrypt the feature vectors of both the modalities obtained before fusion using some encryption algorithms in order to provide heightened security.

Lastly, our system works for only images that are of good quality. For poor quality images, we need to modify our code.

34

Enumerative Bibliography
[1] A. Ross, & A. K. Jain, Information Fusion in Biometrics, Pattern Recognition Letters, 24(132003, 2115-2125.

[2]W. Yunhong, T. Tan, & A. K. Jain, Combining Face and Iris Biometrics for Identity Verification, Proceedings of Fourth International Conference on AVBPA, Guildford, UK, 2003, 805-813.

[3] S. C. Dass, K. Nandakumar, & A. K. Jain, A Principled Approach to Score Level Fusion inMultimodal Biometric Systems, Proc. of Audio- and Video-based Biometric PersonAuthentication (AVBPA), Rye Brook, NY, 2005.

[4] G. Feng, K. Dong, D. Hu, & D. Zhang, When Faces Are Combined with Palmprints: A Novel Biometric Fusion Strategy, International Conference on Bioinformatics and its Applications, Hong Kong, China, 2004, 701-707.

[5] J. Kittler, M. Hatef, R. P. W. Duin, & J. Mates, On combining classifiers, IEEE Transactions on Pattern Analysis and Machine Intelligence, 20(3), pp. 226239, 1998

[6] HunnyMehrotra, AjitaRattani, Phalguni Gupta, Fusion of iris and fingerprint biometric for recognition, Indian Institute of Technology Kanpur, India 208016

[7] Phalguni Gupta, AjitaRattani, HunnyMehrotra, Anil KumarKaushik, Multimodal BiometricsSystem for Efficient Human Recognition, Indian Institute of Technology Kanpur, India 208016.

[8] T. A. Albert, S. Ganesan, Applications of Principal Component Analysis inMultimodal Biometric Fusion System, European Journal of Scientific ResearchISSN 1450-216X Vol.67 No.2 (2012), pp. 248-259 [9] GaganpreetKaur, AkshayGirdhar, ManvjeetKaur, Enhanced Iris Recognition System an IntegratedApproach to Person Identification, International Journal of Computer Applications (0975 8887) Volume 8 No.1, October 2010

35

[10]

SangramBana,

Dr.

DavinderKaur,

Fingerprint

Recognition OF

using

Image

Segmentation,

(IJAEST)

INTERNATIONAL

JOURNAL

ADVANCED

ENGINEERING SCIENCES AND TECHNOLOGIES Vol No. 5, Issue No. 1, 012 023 [11] Masek, L. (2003), Recognition of Human Iris Patterns ForBiometric identification,Thesis Report, The University ofWestern Australia.

[12] George ChellinChandran. J, Dr. Rajesh. R.S, Performance Analysis of Multimodal Biometric SystemAuthentication, IJCSNS International Journal of Computer Science and Network Security, VOL.9 No.3, March 2009.

[13]A. JameerBasha, V. Palanisamy, T. Purusothaman, Fast Multimodal Biometric Approach Using Dynamic Fingerprint Authenticationand Enhanced Iris Features, 978-14244-5967-4/10/$26.00 2010 IEEE

[14]AdemAlpaslan

ALTUN,

Recognition

of

Selected

Fingerprints

and

Iris

FeaturesEnhanced by Curvelet Transform with Artificial Neural Networks.

[15]FetenBESBES, Hanene TRICHIL,Basel SOLAIMAN, Multimodal Biometric System Based on FingerprintIdentification and Iris Recognition.

[16] Li Xiuyan1, Miao Changyun1, Liu Tiegen2, Yuan Chenhu3, Theoretical Analysis and Experimental Study on Multimodal Biometric, 978-1-4577-0860-2/11/$26.00 2011 IEEE.

[17] Mohamad Abdolahi, Majid Mohamadi, Mehdi Jafari, Multimodal Biometric system Fusion Using Fingerprint and Iris with Fuzzy Logic, International Journal of Soft Computing and Engineering (IJSCE) ISSN: 2231-2307, Volume-2, Issue-6, January 2013.

36