DEVELOPMENT OF OPTOELECTRONIC TECHNIQUES FOR RECOGNITION AND ENCRYPTION IN BIOMETRIC BASED SECURITY SYSTEMS
GAURAV VERMA
DEPARTMENT OF PHYSICS
INDIAN INSTITUTE OF TECHNOLOGY DELHI NEW DELHI-110016 INDIA
MAY 2017
© Indian Institute of Technology Delhi (IITD), New Delhi - 2017
DEVELOPMENT OF OPTOELECTRONIC TECHNIQUES FOR RECOGNITION AND ENCRYPTION IN BIOMETRIC BASED SECURITY SYSTEMS
by
GAURAV VERMA 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)
MAY 2017
i CERTIFICATE
This is to certify that the thesis entitled, “Development of Optoelectronic Techniques for Recognition and Encryption in Biometric Based Security Systems ”, being submitted by Mr. Gaurav Verma 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 research work carried out by him under my supervision and guidance. In my opinion the thesis has reached the standards fulfilling the requirements for submission relating to the degree.
The contents of this thesis have not been submitted to any university or institute for the award of any degree or diploma.
Aloka Sinha Professor Department of Physics
Indian Institute of Technology Delhi May, 2017 New Delhi, India-11001
ii ACKNOWLEDGEMENTS
First and foremost, I would like to express my heartfelt gratitude to my supervisor Prof.
Aloka Sinha in the Department of Physics at IIT Delhi, for providing me with the opportunity to do research in the emerging field of biometric related security systems. I am privileged to work under her supervision throughout my research. During my PhD, she has always motivated and encouraged me to develop independent thought process for formulating research problems and their implementation as well as the presentation skills. I heartily acknowledge her thoughtful guidance and inspiration which greatly helped me towards achieving my research objectives.
I would also like to thank my research committee members, Prof. Joby Joseph, Prof.
P. Senthilkumaran and Dr. G.S. Khan, who devoted their precious time to our research progress meetings and for providing their valuable suggestions and advices regarding our research work.
I would like to give my special thanks Prof. M. R. Shenoy, Prof. D. S. Mehta and Dr.
Kedar B Kahre for their insightful discussions on my work. I would like to thank the faculty and the staff of the Physics department, IIT Delhi for their cooperative and supportive attitude and for being of great help. I would like to thank the Nanoscale research facility (NRF), IIT Delhi for providing with sophisticated experimental facility to measure layer thicknesses of the structured patterns of the fingerprint objects. I am also thankful to my lab mates Dr. Nirmala Saini, Mr. Mukesh Kumar Sharma, Mr. Pradeep Kumar, Mrs. Amina Nafees, and Mr. Sourav Patranabish, for their unconditional help and support to accomplish this task. I am ever indebted to them for their distinguished helping nature and fruitful
iii discussions on my research topic. My special appreciation goes to Mr. Avinash Kishore and Mr. Shreyansh Ratnam Khare for their help and advices.
Most importantly, I would like to pay a heartfelt regards to my parents and family for having their trust in me and for being of constant support to me which helped me pursue my academic dreams. My parents were always the source of inspiration and motivation to me.
Words cannot express my deep gratitude for all of the sacrifices that you have done for me. I sincerely hope that their love and blessings will always help me to do better in every step of life. At the end, I would like express appreciation to my beloved wife Lochana for her invaluable companionship, encouragement, inspiration and strong belief in my ability. Her optimistic and enlightening boosts have helped me get through this agonizing period of life in the most positive way. Without her sincere effort, this research work would have never been possible at this stage. To my beloved daughter Prakhya, I would like to express my special thank to the little angel for being in my life and always cheering me up in the moments of ups and downs.
I would like to specially thank Dr. Sunil Verma, Scientific Officer 'G', RRCAT, Indore (M.P.) for providing me with the first exposure towards research. He has been my mentor, my teacher and the supervisor of my M.Tech. project. His guidance and inspiration has always helped me choose the right path in life and to flourish academically.
Above all, I am obliged to the Almighty God, for building confidence and strength in me to continue this research through all the difficulties and tough times, and finally for helping me arrive to a completion of my PhD journey. I have received your kindness, blessings and guidance every day and at every step of my life. I will always keep faith in you. Thank you.
Date (Gaurav Verma)
iv ABSTRACT
In the age of information technology, there is a great demand for confidentiality of information in various areas such as government, financial sectors, banking sectors and private businesses. Thus, the security of information systems has received significant amount of attention in today's world. The most general method used in the area of information security is cryptography. The main measure of security in the cryptographic techniques is based on the keys and the algorithms. These suffer from the issue of confidentiality of the cryptographic keys, which can easily be solved by adopting the biometric based approach due to the distinctive characteristics of a person's biometric traits. Biometric systems are based on an automated process in which physiological or behavioral characteristics of human are utilized for authentication and identification. Biometric systems offer advanced information or data security by the inclusion of the user authentication over traditional schemes such as knowledge-based approach, or possession based approach. Thus, biometric based information security systems have been widely used in practical applications.
This thesis reports the development of different biometric based optical security systems for recognition and encryption. Chapters 2 and 3 utilizes a newly innovated biometric trait i.e. finger knuckle print for recognition. This thesis also addresses the issue of biometric template protection using the concept of cancellability in the field of fingerprint (chapter 4). In these techniques, various digital and optical techniques have been used for feature extraction from the biometrics. Chapters 5 and 6 explore nonlinear image encryption systems in the field of the data security where biometric keys are utilized.
Chapter 1 contains an introduction to biometrics in the field of information security. It presents the brief introduction on different optical and digital techniques used for securing
v information. This chapter presents the description and overview of biometric based recognition system, and the explanation of cancellable biometric. Performance evaluation parameters of the security system are also discussed.
Chapter 2 presents a finger knuckle print recognition system based on advanced correlation filters such as synthetic discriminant function , minimum average correlation energy, minimum variance synthetic discriminant function and optimal tradeoff synthetic discriminant function for personal authentication and verification. The numerical experiments have been carried out to evaluate the performance of the designed filters in terms of the receiver operating characteristic curve, and equal error rate. The obtained results present better discrimination ability between the genuine and the imposter population.
Chapter 3 reports the finger knuckle print recognition system by using wavelet transform and Gabor filters. The wavelet transform is used to decompose the finger knuckle print feature into different frequency subbands, whereas Gabor filters are employed to capture the orientation and frequency. Both the obtained information is utilized for generation of the finger knuckle print template, which is stored for verification. Numerical experiments have been conducted to analyze the performance of the finger knuckle print verification systems in terms of the false rejection ratio, false acceptance ratio, and equal error rate. The proposed scheme is compared with the previous published work in this area.
Chapter 4 presents a new digital holographic based cancellable biometric scheme for personal authentication and verification. The realization of cancellable biometric is based on an optoelectronic experimental approach, in which optically recorded fingerprint hologram of a person is numerically reconstructed. The reconstructed information is utilized to generate a user specific fingerprint features by using a feature extraction process, and then stored for
vi verification. The performance of the proposed system has been evaluated by using the receiver operating characteristic curve, the false acceptance ratio, the false rejection ratio and the equal error rate.
Chapter 5 describes a new method for securing information by using optically generated biometric based keys based on digital holography. For the biometric key generation, the fingerprint information in terms of the amplitude mask and the phase mask of the reconstructed fingerprint image is used. To explore utility of the biometric keys, new optical image encryption system has been formulated based on the phase retrieval algorithm and the double random phase encoding scheme. Numerical simulations have been carried out to verify the feasibility and validity of the proposed technique and compared with the previous reported schemes in this field.
Chapter 6 explores different nonlinear optical image encryption systems in the field of data security based on nonlinear approaches such as using phase-truncated Fourier-transform and logarithms. The introduction of nonlinear transform in the process increases the complexity and security features. As a result, the proposed nonlinear cryptosystem is made immune and resistant against the the amplitude phase-retrieval algorithm based attacks. The proposed scheme provides secure transmission of the encrypted data by the inclusion of biometric features. To measure the robustness of the proposed scheme, the mean square errors and the correlation coefficient are calculated for the decryption process.
Chapter 7 summarizes the key contributions of the research work carried out in the thesis and presents some directions for future extension of this work. Introduction of optical processes and techniques in the future work in this regard may possibly improve the security performance of the biometric systems.
vii
, , , औ ए इ , आ आ औ ए आ , आ आ आ ए आ ए औ ए -आ , आ ए इ , आ औ ए ए आ 2 औ 3 ए ए ( 4) इ , ए औ 5 औ 6
viii - ए ए
1
ए इ औ आ औ , औ
2 , औ , औ औ ए इ आ ए इ ए ए ए ए ए , औ औ आ 3 औ आ ए , औ आ ए ए , ए , औ
ix ए ए ए इ ई
4 औ ए ए आ ए इ आ , ए ए ए ए ए , औ ए , , औ
5 आ आ ए ए ई ए, आ औ ए, ई इ ए ए औ ए आ औ ए औ इ
6 - इ ए , -
x - आ - - औ इ - औ , - इ आ - ए आ औ ए ए, औ ए
7, ए ए औ इ ए - इ औ
xi TABLE OF CONTENTS
Certificate i
Acknowledgements ii
Abstract iv
Table of Contents xi
List of Figures xvii
List of Tables xxiv
List of Abbreviations xxv
CHAPTER 1: INTRODUCTION 1–27 1.1 Biometrics 3
1.1.1 Biometric system design and its mode of operation 7
1.2 Cancellable biometrics 10
1.3 Feature extraction techniques 12
1.3.1 Advanced correlation filter based techniques 12
1.3.2 Multiresolution scheme 14
1.3.2.1 Discrete wavelet transform 15
1.3.2.2 Gabor filter 15
1.3.3 Digital holography 16
1.3.4 Principle component analysis (PCA) 17
1.4 Security aspect and key management of an optical cryptosystem 18
1.4.1 Double random phase encoding (DRPE) scheme 19 1.4.2 Phase-truncated Fourier transforms (PTFT) based encryption
xii
scheme 21
1.5 Performance evaluation 23
1.6 Biometric databases 26
1.6.1 Finger knuckle print database (Poly U FKP database) 26
1.6.2 Face databases (Yale B database) 26
1.6.3 Fingerprint database (FVC 2004 database) 27
1.7 Conclusion 27
CHAPTER 2: FINGER KNUCKLE PRINT BASED VERIFICATION USING ADVANCED CORRELATION FILTER 28–53 2.1 Introduction 28
2.2 Advanced correlation filters (ACFs) 30
2.2.1 Synthetic discriminant function (SDF) filter 31
2.2.2 Minimum average correlation energy (MACE) filter 32
2.2.3 Minimum variance synthetic discriminant function (MVSDF) filter 34
2.2.4 Optimal trade-off filters 35
2.3 The region of interest extraction process of a finger knuckle print 35
2.4 Recognition process of FKP using ACF 37
2.5 Performance analysis 38
2.6 Results and discussion 39
2.6.1 SDF filter 39
2.6.2 MACE filter 42
2.6.3 MVSDF filter 47
xiii
2.6.4 OTSDF filter 49
2.7 Conclusion 52
CHAPTER 3: FINGER KNUCKLE PRINT RECOGNITION BASED ON WAVELET AND GABOR FILTERING 54–79 3.1 Introduction 54
3.2 Methodology 56
3.2.1 Wavelet based approach 56
3.2.1.1 Selection of wavelet 61
3.2.2 Gabor filter 63
3.3 Proposed techniques 66
3.3.1 Proposed scheme for feature extraction 67
3.3.1.1 Feature extraction of FKP using wavelet 67
3.3.1.2 Feature extraction of FKP using Gabor filtering 68
3.4 Performance evaluation and matching parameter 69
3.5 Results and discussions 71
3.5.1 Experiment 1 71
3.5.2 Experiment 2 73
3.6 Conclusion 79
CHAPTER 4: DIGITAL HOLOGRAPHIC BASED CANCELLABLE BIOMETRIC FOR PERSONAL AUTHENTICATION 80–104 4.1 Introduction 80
4.2 Proposed scheme 83
xiv
4.2.1 Recording of the fingerprint hologram 84
4.2.2 Reconstruction process 86
4.2.3 Feature extraction 89
4.2.4 Evaluation of the cancellable approach 90
4.3 Results and discussion 98
4.3.1 Verification results 99
4.3.2 Security analysis 102
4.4 Conclusion 104
CHAPTER 5: SECURING INFORMATION USING OPTICALLY GENERATED BIOMETRIC KEYS 105–131 5.1 Introduction 105
5.2 Proposed scheme 107
5.2.1 Generation of the biometric keys 109
5.2.2 Features of the fingerprint biometric keys 112
5.2.2.1 Randomness 112
5.2.2.2 Discrimination ability 113
5.3 Biometric keys based optical encryption system 117
5.3.1 Encryption scheme 117
5.3.2 Decryption process 121
5.4 Numerical experiment and analysis 123
5.4.1 Results 123
5.4.2 Robustness of the biometric key verification 125
5.4.3 Security analysis 128
xv
5.4.4 Comparative analysis 130
5.5 Conclusion 131
CHAPTER 6: NEW NONLINEAR OPTICAL IMAGE ENCRYPTION SYSTEMS 132–161 6.1 Introduction 132
6.2 Nonlinear optical cryptosystem free from amplitude-phase retrieval attacks 134
6.2.1 Proposed scheme 134
6.2.2 Attack analysis 138
6.2.2.1 Special attack 138
6.2.2.2 Known plaintext attack 139
6.2.2.3 Chosen plaintext attack 140
6.2.3 Results and discussion 141
6.3 Optical image encryption system using nonlinear approach based on biometric authentication 146
6.3.1 Proposed scheme 146
6.3.1.1 Encryption process 148
6.3.1.2 Generation of the face biometric based phase mask key 149 6.3.1.3 Decryption process 150
6.3.2 Simulation results and discussion 153
6.3.2.1 Results 153
6.3.2.2 Robustness of the AM key 156
xvi
6.3.2.3 Security analysis 158
6.4 Conclusion 161
CHAPTER 7: CONCLUSION AND FUTURE WORK 162–168 7.1 Conclusion 162
7.2 Future work 166
REFERENCES 169
LIST OF PUBLICATIONS 187
AUTHOR’S BIOGRAPHY 189
xvii LIST OF FIGURES
Fig. 1.1 Biometric features (physiological and behavioral) (a) Fingerprint (b) Face (c) FKP (d) Palmprint (e) Keystroke dynamics (f) Ear (g) Iris (h) Voice
pattern (i) Signature. 6
Fig. 1.2 Enrolment process of a person in a biometric system. 8
Fig. 1.3 Verification process of a biometric system. 8
Fig. 1.4 Identification process of a biometric system. 9
Fig. 1.5 DRPE scheme (a) encryption (b) decryption. 20
Fig. 1.6 PTFT based cryptosystem (a) encryption (b) decryption. 21
Fig. 1.7 Performance behaviour of a biometric system (a) Genuine and imposter distribution (b) Trade-off between FAR and FRR. 24
Fig. 2.1 (a) Original FKP image (b) Extracted ROI of FKP image. 37
Fig. 2.2 Block diagram of correlation process. 38
Fig. 2.3 (a) FKP training images (b) Representation of SDF filter in space domain. 39-40 Fig. 2.4 The SDF correlation output (a) Genuine person (b) Imposter person. 40
Fig. 2.5 The graph plotted between number of subjects and average of maximum peak values for SDF filter. 41
Fig. 2.6 MACE filter output for a person (a) Genuine (b) Imposter. 42
Fig. 2.7 (a) Graph of PSR performance for subject 1 (b) Graph of PSR performance for subject 2. 43 Fig. 2.8 (a) Graph between number of images and normalized absolute difference of the PSR for subject 1 (b) Graph between number of images and normalized
xviii
absolute difference of the PSR for subject 2. 45
Fig. 2.9 (a) Noisy image (b) Correlation output. 46
Fig. 2.10 (a) Original FKP image (b) 20 dB SNR. 48
Fig. 2.11 The MVSDF correlation output (a) Genuine person (b) Imposter person. 48
Fig. 2.12 The graph of average maximum values of correlation output and subject numbers for MVSDF filter of 20 dB SNR. 49
Fig. 2.13 The OTSDF correlation output (a) α =0 (b) α =1. 50
Fig. 2.14 The OTSDF correlation output for genuine person (a) α =0.33 (b) α =0.67. 50 Fig. 2.15 The OTSDF correlation output for imposter person (c) α =0.33 (d) α =0.67. 51
Fig. 3.1 Presentation of wavelet. 57
Fig. 3.2 DWT implementation using filters banks. 59
Fig. 3.3 Representation of a one level two dimensional DWT decomposition of FKP image. 60
Fig. 3.4 Wavelet decomposition level (a) 1st level (b) 2nd level (c) 3rd level. 61
Fig. 3.5 Several different families of wavelets. 63
Fig. 3.6 The six orientation of Gabor representation. 65
Fig. 3.7 Block diagram of proposed FKP verification system. 66
Fig. 3.8 (a) Original FKP image (b) ROI of FKP image. 67
Fig. 3.9 The subbands representation of FKP images at all three-decomposition level for the Haar wavelet. 68
Fig. 3.10 Six FKP orientation and frequencies of Gabor filtered FKP image. 69
Fig. 3.11 ROC curve for Haar wavelet. 72
xix Fig. 3.12 Graph between population density and normalized Euclidean Distance. 74
Fig. 3.13 ROC curve of combined scheme for Haar wavelet. 74 Fig. 4.1 Flow diagram of the proposed cancellable fingerprint biometric scheme. 83
Fig. 4.2 Experimental set up for recording digital hologram of fingerprint image: SF:
spatial filtering, Ms: mirrors, BSs: beam splitters. 84 Fig. 4.3 Digital hologram of the fingerprint. 86 Fig. 4.4 Schematic diagram expressing the relation between the object plane, the recording plane and the reconstruction plane. 87 Fig. 4.5 Reconstructed features from Fig. 4.3 at d = 0.29 m 89 Fig. 4.6 Extracted features from Fig. 4.5 at d = 0.29 m (a) Amplitude distribution (b) Generated phase mask (PM). 90 Fig. 4.7 Visualization of the multiple reconstruction planes by variation of the
reconstruction distance (Δd ). 91 Fig. 4.8 Graph between the maximum value of correlation output and variation of the
reconstruction distance. 92 Fig. 4.9 Reconstructed feature from the fingerprint hologram (see Fig. 4.3) at the
reconstrction distance (a) Δd = 0.27 m (b) d + Δd = 0.31 m
(c) d + 2Δd = 0.33 m. 93 Fig. 4.10 Correlation output between ‘F1’, ‘F2’, ‘F3’ and ‘F4’: (a) (F1, F1) (b) (F1, F2) (c) (F1,F3) (d) (F1, F4)(e) (F2,F2)(f) (F2, F1) (g) (F2,F3) (h) (F2, F4) (i) (F3,F3) (j) (F3, F1)(k) (F3,F2)(l) (F3, F4) (m) (F4,F4)(n) (F4, F1) (o) (F4, F2) (p) (F4, F3).
94-96 Fig. 4.11 Schematic diagram expressing discrimination performance by plotting the
xx maximum value of peak for all the correlation output between ‘F1’, ‘F2’, ‘F3’
and ‘F4’, which is obtained at the reconstruction distance by varying Δd. 97
Fig. 4.12 Correlation output for a person: (a) Genuine (b) imposter. 100
Fig. 4.13 Population density discrimination graph. 101
Fig. 4.14 ROC curve. 102
Fig. 5.1 Flowchart for the biometric keys generation. 108
Fig. 5.2 Recorded digital hologram of the fingerprint biometric. 110
Fig. 5.3 Reconstructed fingerprint images (a) amplitude-contrast (b) phase-contrast. 111 Fig. 5.4 Generated fingerprint phase mask (PM). 112
Fig. 5.5 Results of the generated biometric keys at λ=632.8 nm, d=29 cm. (a) and (e) fingerprint holograms of different persons; (b–c) and (f–g) both the amplitude contrast and the phase-contrast images; (d) and (h) its obtained PM key, respectively. 114
Fig. 5.6 Simulation results for correlation output: Autocorrelation peaks for (a) PM1 key (b) PM2 key (c) PM3 key. 115
Fig. 5.7 Simulation results for correlation output: Cross-correlation peaks for (a) (‘PM1’, ‘PM2’) (b) (‘PM1’, ‘PM3’) (c) (‘PM2’, ‘PM1’) (d) (PM2’, ‘PM3’) (e) (‘PM3’, ‘PM1’) (f) (‘PM3’, ‘PM2’). 116
Fig. 5.8 Flow chart of the proposed scheme. 118
Fig. 5.9 Impact of the number of iterations (a) MSE convergence graph (b) Correlation coefficient (CC) graph. 120 Fig. 5.10 The parameter values on recovered image with number of iterations: (a) MSE
xxi = 0.321015, CC = 0.868, Iteration no = 10; (b) MSE = 0.321011, CC = 0.988, Iteration no = 100; (c) MSE = 0.321011, CC = 0.995, Iteration no =
300; (d) MSE = 0.321011, CC = 0.996, Iteration no =403. 120
Fig. 5.11 DRPE scheme for encryption. 121
Fig. 5.12 DRPE scheme for decryption. 122
Fig. 5.13 Flow chart of the decryption process. 122
Fig. 5.14 Optical setup for decryption, f: focal length, L: lenses, CCD: Charge coupled device, SLM: Spatial light modulator. 123
Fig. 5.15 (a) Input image (b) Fingerprint AM key (c) Fingerprint PM key (d) RPM key (DRPE). 124
Fig. 5.16 Decryption results (a) Initial ciphertext (C1) (b) Ciphertext (E) (c) Using all correct keys (d) keys in wrong positions (e) Different fingerprint PM key in place of the original fingerprint PM key (f) No keys. 125
Fig. 5.17 MSE curve with respect to the variation of the reconstruction parameters (a) Distance d = 0.29 ± 0.02 m, λ = 632.8 nm (b) Wavelength λ = 632.8 ± 10 nm, d = 0.29 m. 126
Fig. 5.18 CC curve with respect to the variation of the reconstruction parameters (a) Distance d = 0.29 ± 0.02 m, λ = 632.8 nm (b) Wavelength λ = 632.8 ± 10 nm, d = 0.29 m. 127
Fig. 5.19 Decrypted results using the arbitrarily generated fingerprint biometric keys (a) MSE = 0.1194, CC = -0.0016 at d = 0.25 m, λ = 632.8 nm (b) MSE = 0.1183, CC = 0.0027 at d = 0.29 m, λ = 592.8 nm. 128 Fig. 5.20 (a) Input image (b) Initial ciphertext (C1) (c) Encrypted (E) (d) KPA Attack
xxii result. 129 Fig. 6.1 Flow chart of the proposed scheme (a) Encryption (b) Decryption. 135 Fig. 6.2 Proposed optical setup for the decrypting input image, L: lenses, CCD:
Charge coupled device, SLM: Spatial light modulator, PC: Personal
computer. 137 Fig. 6.3 (a) Input image (b) the decryption key (D1) (c) the decryption key (D2)
(d) the decryption key (D3) (e) Ciphertext. 141 Fig. 6.4 Decrypted results (a) True decryption (b) No Keys (c) Incorrect RPMs
Keys (d) Correct D3 while D1 and D2 are put in the wrong position (e) Correct D2 while D1 and D3 are put in the wrong position (f) Correct D1 while D2 and D3 are put in the wrong position. 142 Fig. 6.5 Attack results for 50 iteration in first step (a) The MSE between E2(m,n) and
E'2(m,n) (b) Approximate amplitude. 143
Fig. 6.6 Attack results for 500 iteration number in second step (a) The MSE between E1(u,v) and E'1(u,v) (b) Approximate amplitude. 143 Fig. 6.7 Attack results for 1000 iteration number in third step (a) The MSE
convergence (b) Approximate amplitude. 144 Fig. 6.8 Known plaintext attack results (a) Plot of MSE versus iteration numer when decryption key K3 is obatained (b) Plot of MSE versus iteration numer when decryption key K2 is generated (c) Plot of MSE versus
iteration numer when decryption key K1 is obatained (d) Final, decrypted image. 145 Fig. 6.9 Flow chart of the proposed scheme: (a) encryption scheme, (b)
xxiii decryption process. 147 Fig. 6.10 Proposed optoelectronic experimental setup for decryption: f: focal length, L: lenses, CCD: Charge coupled device, SLM: Spatial light
modulator. 152 Fig. 6.11 (a) Input image (b) RPM1 (c) RPM2 (d) Secret key (D1) (e) Ciphertext. 153 Fig. 6.12 Key generation: (a) Face image (b) FBPM key. 154 Fig. 6.13 Decryption results :(a) Decrypted image using all correct keys (b)
Decrypted image using a wrong FBPM key (c) Decrypted image using
an incorrect AM (d) Decrypted image without using secret key (e) Decrypted image using an incorrect secret key. 155
Fig. 6.14 Curve with respect to the variation of the distribution of k values in the
amplitude mask (a) MSE curve (b) CC curve. 157 Fig. 6.15 KPA results: (a) Plot of MSE and number of iteration for the obtaining
secret key (b) Plot of MSE and number of iteration for the second key. 160 Fig. 6.16 A retreived image after 500 iterations. 160
xxiv LIST OF TABLES
Table 2.1 EER calculation for the MACE filter.
Table 2.2 Comparison with earlier techniques.
Table 2.3 PSR values for different values of ‘α’.
Table 3.1 The Gabor filters parameter.
Table 3.2 Frequency subbands level of FKP images and its resolution.
Table 3.3 Recognition rate for the FKP verification at different decomposition level.
Table 3.4 Recognition rate of combined scheme for the FKP verification.
Table 3.5 Comparative evaluation of proposed techniques with existing methods for FKP verification.
Table 3.6 Comparison of proposed techniques with wavelet based methods for different biometrics verification.
xxv LIST OF ABBREVIATIONS
ACE ACF AM ATM CC CCA CCD DRPE DWT ED EER FAR FBPM FKP FRR FT IFT KPA MACE MSE MSF
AVERAGE CORRELATION ENERGY ADVANCED CORRELATION FILTER AMPLITUDE MASK
AUTOMATED TELLER MACHINE CORRELATION COEFFICIENT CHOSEN CIPHERTEXT ATTACK CHARGE COUPLED DEVICE
DOUBLE RANDOM PHASE ENCODING DISCRETE WAVELET TRANSFORM EUCLIDIAN DISTANCE
EQUAL ERROR RATE
FALSE ACCEPTANCE RATIO
FACE BIOMETRIC BASED PHASE MASK FINGER KNUCKLE PRINT
FALSE REJECTION RATIO FOURIER TRANSFORM
INVERSE FOURIER TRANSFORM KNOWN PLAINTEXT ATTACK
MINIMUM AVERAGE CORRELATION ENERGY MEAN SQUARE ERROR
MATCHED SPATIAL FILTER
xxvi MVSDF
ONV OTSDF PCA PIN PM PR PST PT PTFT PSR ROC ROI RPM RSA SDF SLM SNR STFT UID WT
MINIMUM VARIANCE SYNTHETIC DISCRIMINANT FUNCTION OUTPUT NOISE VARIANCE
OPTIMAL TRADEOFF SYNTHETIC DISCRIMINATION FUNCTION PRINCIPLE COMPONENT ANALYSIS
PERSONAL IDENTIFICATION NUMBER PHASE MASK
PHASE RESERVED
PIXEL SCRAMBLING TRANSFORM PHASE TRUNCATED
PHASE TRUNCATED FOURIER TRANSFORM PEAK TO SIDE LOBE RATIO
RECEIVER OPERATING CHARACTERISTICS REGION OF INTEREST
RANDOM PHASE MASK
RIVEST- SHAMIR- ADLEMAN
SYNTHETIC DISCRIMINANT FUNCTION SPATIAL LIGHT MODULATOR
SIGNAL TO NOISE RATIO
SHORT TIME FOURIER TRANSFORM UNIQUE IDENTIFICATION
WAVELET TRANSFORM