INVESTIGATIONS ON BIOMETRIC BASED INFORMATION SECURITY SYSTEMS
by
NIRMALA SAINI Department of Physics
Submitted
in fulfillment of the requirements of the degree of Doctor of Philosophy
to the
INDIAN INSTITUTE OF TECHNOLOGY DELHI NEW DELHI-110016 INDIA
JANUARY 2013
DEDICATED TO
DEDICATED TO
DEDICATED TO
DEDICATED TO
My Ma and Bauji
My Ma and Bauji My Ma and Bauji
My Ma and Bauji
i
CERTIFICATE
This is to certify that the thesis entitled “INVESTIGATIONS ON BIOMETRIC BASED INFORMATION SECURITY SYSTEMS” being submitted by Ms. Nirmala Saini to the Department of Physics, Indian Institute of Technology Delhi for the award of the degree of Doctor of Philosophy. This thesis is a record of bona-fide work carried out by her under my guidance and supervision. In my opinion the thesis has reached the standards fulfilling the requirements for submission relating to the degree.
The results contained in this thesis have not been submitted to any other University/Institute for the award of any degree or diploma.
ALOKA SINHA ASSOCIATE PROFESSOR DEPARTMENT OF PHYSICS
INDIAN INSTITUTE OF TECHNOLOGY DELHI JANUARY 2013 HAUZ KHAS 110016, NEW DELHI, INDIA
ii
ACKNOWELEDGEMENTS
First and foremost, I would like to express my deepest gratitude to my parents for their unconditional love, support and blessing. I believe their blessing will be always with me throughout my life as a source of inspiration. It was the positive attitude of my father which has always given me the strength to pursue my dreams. I feel proud to dedicate this thesis and all the good things of my life to them.
I feel honored to express my sincere gratitude to my supervisor Dr. Aloka Sinha for providing me an opportunity to work in the challenging and emerging field of biometric security systems. Her constant motivation, guidance, encouragement helped me for the successful completion of this thesis and to achieve my research goal. Discussions with her have been extremely helpful in improving my understanding of the subject.
I am thankful to my research committee members Prof. P. Senthilkumaran and Prof.
D.S. Mehta for their valuable suggestions and comments which surely improved the presentation of my thesis. Thanks are due to the faculty and staff of Department of Physics for providing a cooperative and supportive environment in the department. I would like to specially thank to Prof. H.C. Gupta and Prof. Joby Joseph for their help during the stay at IIT.
I am also thankful to my colleagues Dr. Narendra Singh, Mrs. Priya Sinha, Mr.
Gaurav Verma, Ms. Swarnalatha and Mr. Manoj Prasad for their cordial support during my stay in the lab. My special thank to Mr. Amioy Kumar, Dr. Debjani Bhattacharya and Ms.
Swapna Mishra for their help and discussions on Matlab programming and simulation. I would like to thank to Dr. Trilok Singh, Mr. Rajuram Parihar, Mr. Sonu, Mr. Karan, Mr.
Rohan Saini and my husband Mr. Kailash Chand Saini for volunteering for data collection of face database.
I would like to thank my friends Ms. Swapna Mishra, Dr. Tulsi Anna, Ms. Anupama Sharma, Dr. Trilok Singh, Dr. Alok Jha, Dr. Vandana Gupta, Mr. Rajuram Parihar, Dr.
iii
Nandan Bisht, Dr. Hema Bora, Ms. Usha Parihar, Ms. Shalini Singh, Ms. Ruchi Agrawal for helping me in various ways at each step of my life. I gained a lot from them, through their personal and scholarly interactions and their suggestions at various point of my research programme.
I would like to express my indebtedness to my whole family for their motivation, support, affection, help and love. Without their sincere efforts, it would not have been possible for me to continue my studies. My heartfelt thanks to my eldest sister Ms. Pushpa Devi and her husband Mr. Kailash Chand Saini for loving taking the role of my parents. My special thanks to my brother Mr. Pawan Kumar Saini for raising a burning desire in me to do something worthwhile in life. I am thankful to my cousin sister Ms. Sugana Devi and her husband Mr. B. L. Saini for their strong support throughout the tenure of Ph.D. I am very thankful to my nephew Mr. Rajesh Saini for his invaluable support during my stay at Delhi.
I am thankful to Mr. Kailash Chand Saini, my husband without whose support this work would not have been possible. He not only supported me emotionally throughout my Ph.D. but also accompanied me during late night stays for various experiments. I am thankful to him for critical reading of my thesis and his valuable suggestions. His encouragement kept me motivated even during the most difficult periods. My special thanks to my little angel Titiksha for being in my life. Her sweet smile spreads a new enthusiasm and energy in my heart.
Last but not least, I would like to thank my Guruji (Ashutoshji Maharaj) for making me realized that almighty God will be always with me to show the right path in life and help me flourish academically and spirituality.
Nirmala Saini
iv
ABSTRACT
Securing information is of paramount concern for the modern networked society. Information security includes the protection of user data and user authentication along with ensuring the privacy and integrity of information. These issues of information security can be solved by using biometrics and cryptography. Biometrics is a physiological and behavioural characteristic of a human being that can be used to identify and verify the identity of a person. Biometrics provides greater security and convenience than traditional authentication methods since they are associated with the user. Biometric technologies are widely used for physical access and for security of the systems.
Cryptography is concerned with the protection of information using mathematical formulations. Cryptographic systems are used to provide confidentiality, data integrity, entity authentication and data origin authentication. A combination of biometrics and cryptography has the potential to provide higher security and authentication to the system. Optical security systems are also emerging techniques for information security. They are relatively new, but they offer a promising approach to secure information. These systems have the distinct advantage of processing complex data in parallel at great speeds.
The thesis reports different biometric security systems for recognition and security purposes. These techniques are basically optics based biometric security systems. In these techniques different transforms have been used to extract the features from the biometrics.
The thesis also contains the key management of security systems using public key encryption and biometrics.
Chapter 1 consists of an over all introduction to the different areas of research in information security. An introduction to different biometric systems for encryption as well as recognition
v
process has been given. A brief discussion on cancellable biometrics has also been included.
Various biohashing and multimodal biometric techniques have also been introduced. It also includes an introduction to different biometric techniques using soft biometrics. Different parameters of measurement have also been introduced. This chapter provides a brief discussion on various optical and digital techniques used for information security. A brief introduction about the issue of key management of a symmetric key system has been given.
Public key encryption has also been described. Discussion about the error control codes for key management has been included.
Chapter 2 includes a biometric recognition system based on an optics based biohashing technique. In a biohashing system, a biohash code is generated by combining the features of the biometric with the orthonormalized random numbers. The novelty of the technique lies in the fact that the optics based joint transform correlator (JTC) has been used for extraction of the specific feature of the biometric for biohashing. In the enrolment process, a biohash code has been generated by using a face image of the enrolled person. In the verification process, this stored biohash code is matched with the verification codes for recognition purposes.
Experimental as well as simulation results have been given to validate the proposed technique. In the experimental setup of the JTC, spatial light modulator has been used to display the input images. The correlation peaks obtained by JTC have been captured by using a charged coupled device (CCD) camera. Normalized Hamming distance has been used as the discrimination factor. The study of the variation of the normalized Hamming distance with the density of the population for varying dimension of the feature matrix has been undertaken. The performance of the technique has been evaluated by using the receiver operating characteristic (ROC) curve. A comparative study of the proposed technique with the techniques existing in literature has also been carried out.
vi
Chapter 3 describes a new biohashing technique in which soft biometrics of a person has been used to further improve the results obtained by using the earlier proposed biohashing technique. Soft biometric traits are those which are not sufficient to establish the identity of a person such as gender, eye colour, ethnicity, height, weight, face ratio and length etc. In this technique, the biohash code has been generated by the discretization of the biometric feature with orthonormalized random numbers. In the enrolment process, biohash code of the target face images has been integrated with the different soft biometrics of the same person. The obtained code has been stored for verification. In the verification process, the biohash code of the face image to be verified is again fused with the soft biometric of the person. The obtained code is matched with the stored code of the target. In order to fuse the soft biometrics with the biohash codes weighted sum fusion method has been used. Experimental as well as simulation results have been presented to validate the proposed technique. Three types of cases have been studied to see the contribution of tokenised random number in the biohashing. A detailed study has been carried out to find out the optimum values of the weighting factor for the different biometrics. The ROC curve and the equal error rate (EER) have been used to evaluate the performance of the technique.
Chapter 4 explores different multimodal biometric systems in which multiple biometric traits of a user have been used for recognition. The features of the multiple images, face and palmprint images have been extracted by using the Gabor-Wigner transform [24] and Gabor filtered Wigner transform. The particle swarm optimization (PSO) technique has been used for two different purposes. The first purpose is to optimise the parameters of the Gabor filter.
The second purpose is to select the dominant features from the feature vectors. Different levels of fusion have been used to integrate the features of the different biometrics. A detailed study has been carried out to show the results obtained by different unimodal and multimodal
vii
systems. The performance of the system has been evaluated in terms of the ROC curve and the EER.
Chapter 5 deals with the key management technique in which the key of the optical encryption method has been secured by using public key encryption. In the proposed technique, an input image has been encrypted by using the double random phase encoding method using extended fractional Fourier transform. The key of the encryption process have been encoded by using the Rivest–Shamir–Adleman (RSA) public key encryption algorithm.
The encoded key has then been transmitted to the receiver side along with the encrypted image. In the decryption process, the encoded key has been decrypted using the secret key and then the retrieved key parameters have been used to decrypt the encrypted image. The problem associated with the management and transmission of the key has been eliminated by using public key encryption. Numerical simulation has been carried out to verify the proposed technique.
Chapter 6 presents the technique in which the key of the optical encryption method has been secured by using biometrics of the enrolled person. In this technique, the key of the double random phase encoding technique has been linked to the biometric of the enrolled person by using a look up table. This lookup table has been formed by using the encryption key and the biometric template. The biometric template has been formed by using the fingerprint of the user in the log polar domain. The linking algorithm not only secures the encryption key but also authenticates the input image. The look up table has been transmitted to the receiver side, instead of the transmission of the key. The main advantage of this method is its capability to retrieve the same key in the decryption process by using the live fingerprint of the enrolled person. Numerical experiments have been carried out on Matlab platform to validate the
viii
proposed technique. The signal to noise ratio and mean square error has been calculated in order to support the proposed technique.
Chapter 7 includes the technique in which the key of the optical encryption method has been encoded by using the Bose–Chaudhuri–Hocquenghem (BCH) code. In this technique, the encoded key has been linked to the features extracted from the biometrics of the user using exclusive-OR (XOR) operation. In order to improve the performance, shuffling key has been used to shuffle the encoded key. The shuffling key has then been encoded using RSA public key encryption to further enhance the security of the system. The encoded shuffling key and the data obtained by XOR operation have been stored in a token. For the feature extraction Gabor filter has been used. The proposed technique has been demonstrated on face and fingerprint databases. In the decryption process, the key retrieval is possible only in the simultaneous presence of the token and the biometrics of the user. Computational experiments have been carried out to validate the proposed technique. The performance of the system has been evaluated in terms of the EER.
Chapter 8 includes conclusion and contribution of the thesis and scope for future work in the area of information security using biometrics. This includes a new emerging biometric finger knuckle print for biometric recognition systems and biometric security systems.
ix
TABLE OF CONTENTS
Certificate i
Acknowledgements ii
Abstract iv
Table of contents ix
List of figures xv
List of tables xxii
List of abbreviations xxiv
CHAPTER 1: INTRODUCTION 1-33 1.1Biometric security system 3
1.2Cancellable biometrics 6
1.2.1 Biohashing technique 6
1.3Soft biometrics 7
1.4Multimodal biometric system 8
1.5Different feature extraction techniques 11
1.5.1 Principle component analysis 12
1.5.2 Gabor transform 13
1.5.3 Wigner distribution function 14
1.5.4 Gabor – Wigner transform 15
1.5.5 Gabor filter 15
1.5.6 Gabor filtered Wigner transform 16
1.5.7 Log polar transform 16
1.6Feature selection using particle swarm optimization 17
1.7Databases 19
x
1.7.1 Self generated face databases 19
1.7.2 Publicly available databases 20
1.7.2.1Face databases 20
1.7.2.2Palm print databases 21
1.7.2.3Finger print databases 21
1.8Performance evaluation 22
1.9 Key management of security systems 24
1.9.1 Double random phase encoding 24
1.9.2 Double random phase in fractional Fourier domain 27
1.9.3 Double random phase encoding in extended fractional Fourier domain 27
1.10 Chaos function 28
1.11 Public key encryption 29
1.12 Key management using biometrics 30
1.13 Error correcting codes 31
1.13.1 BCH code 32
1.14 Conclusion 32
CHAPTER 2: OPTICS BASED BIOHASHING TECHNIQUE USING JOINT TRANSFORM CORRELATOR 34-55 2.1. Introduction 34
2.2. Proposed technique 36
2.2.1 Feature extraction from the biometric image 37
2.2.2 Biohashing 38
2.3. Databases 39
2.4. Simulation results 40
xi
2.5. Experimental setup 43
2.6. Experimental results 43
2.6.1 Enrolment process 44
2.6.2 Verification process 45
2.7. Performance evaluation 46
2.8. Comparative studies 52
2.8.1 Biohashing using PCA 52
2.9. Conclusion 54
CHAPTER 3: SOFT BIOMETRICS IN CONJUNCTION WITH OPTICS BASED BIOHASHING 56-73 3.1. Introduction 56
3.2. Proposed algorithm 57
3.2.1 Biometric image acquisition 59
3.2.2 Enrolment process 60
3.2.2.1 Fusion of different biometrics 61
3.2.3 Verification process 62
3.3. Simulation results 64
3.4. Experimental setup 65
3.5. Experimental results 66
3.6. Performance evaluation 67
3.7. Conclusion 72
xii
CHAPTER 4: FACE AND PALMPRINT MULTIMODAL BIOMETRIC SYSTEM USING GABOR FILTERED WIGNER TRANSFORM AS FEATURE EXTRACTION
74-98
4.1. Introduction 74
4.2. Proposed algorithm 76
4.2.1 Unimodal biometric system 77
4.2.1.1 Unimodal biometric system based on GWT 77
4.2.1.2 Unimodal biometric system based on GFWT 78
4.2.2 Multimodal system based on feature level fusion 79
4.2.2.1 Feature level fusion based on GWT 79
4.2.2.2 Feature level fusion based on GFWT 81
4.2.3 Multimodal system based on matching score level fusion 82
4.2.4 Hybrid multimodal system 84
4.3. Feature selection using particle swarm optimization 85
4.4. Databases 86
4.5. Experimental results and analysis 87
4.6. Conclusion 98
CHAPTER 5: KEY MANAGEMENT OF THE DOUBLE RANDOM PHASE ENCODING METHOD USING PUBLIC KEY ENCRYPTION 99-112 5.1. Introduction 99
5.2. RSA public-key cryptosystem 101
5.3. Proposed technique 102
5.3.1 Encryption process 103
5.3.2 Decryption process 106
xiii
5.4. Computer simulation 107
5.5 Security analysis 108
5.6 Robustness analysis 109
5.7. Result and discussion 111
5.8. Conclusion 112
CHAPTER 6: KEY MANAGEMENT OF THE DOUBLE RANDOM PHASE ENCODING METHOD USING BIOMETRICS 113-133 6.1. Introduction 113
6.2. Proposed technique 115
6.2.1 Encryption process 115
6.2.1.1 Design of enrollment model (Stage E-1) 116
6.2.1.2 Key linking algorithm (Stage E-2) 117
6.2.1.3 Encryption of the image (Stage E-3) 117
6.2.2 Decryption process 118
6.2.2.1 Design of Verification model (Stage V-1) 119
6.2.2.2 Key retrieval algorithm (Stage V-2) 120
6.2.2.3 Decryption of encrypted image (Stage V-3) 120
6.3. Simulation results 121
6.4. Algorithm analysis 124
6.4.1 Robustness analysis 124
6.4.2 Rotation and scaling analysis 125
6.4.3Analysis of key retrieval from fingerprints with the same appearance 131
6.5. Conclusion 132
xiv
CHAPTER 7: BIOMETRICS BASED KEY MANAGEMENT OF DOUBLE RANDOM PHASE ENCODING SCHEME USING ERROR CONTROL CODES 134-153
7.1. Introduction 134
7.2. Proposed technique 136
7.2.1 Encryption process 136
7.2.1.1 Feature extraction from fingerprint database 136
7.2.1.2 Feature extraction from face database 139
7.2.1.3 Encryption of input image 139
7.2.1.4 Linking of the encoded key to the biometrics 142
7.2.2 Decryption process 142
7.2.2.1 Feature extraction and key retrieval 143
7.2.2.2 Decoding of the key and decryption of the encrypted image 143
7.3. Numerical experiments 144
7.4. Experimental results 147
7.5. Security analysis 151
7.6. Discussion 152
7.7. Conclusion 153
CHAPTER 8: CONCLUSION AND THE FUTURE SCOPE FOR FURTHER STUDIES 154-157 8.1 Summary of important contributions 154
8.2 Scope for future studies 157
REFERENCES 158 LIST OF PUBLICATIONS
AUTHOR’S BIOGRAPHY
