Selected problems of biometry

Monographic approach to selected problems of biometrics - science dealing with determining or confirming the identity of people based on their appearance, physical characteristics and behavior. The basics of automatic object recognition are given including, among others selection and extraction of features and various data classification methods, as well as probabilistic methods for the recognition and classification of vector sequences. The basic biometric methods were also described, including the use of physical characteristics (automatic fingerprint analysis, automatic iris recognition, automatic face recognition and recognition based on DNA analysis) and behavioral characteristics (voice recognition, recognition based on style analysis and signature dynamics) ).

---

Table of Contents:
Introduction 7
Part I Automatic recognition of objects
1. Introduction 13
2. Features and spaces of features 16
2.1. Selection of features 17
2.1.1. Selection of features by maximizing the efficiency of classification 18
2.1.2. Selection of features by maximizing class separability 19
2.1.3. Unattended selection of features 22
2.2. Extraction of features 22
2.2.1. Unattended extraction of features - PCA method
2.2.2. Supervised feature extraction - LDA method
2.2.3. Other methods of extraction of features 31
3. Data classification 33
3.1. Low-level classification 35
3.1.1. Classification of data by k- NN 37
3.1.2. Distance measurements used in the classification of 40
3.2. Classification by division of feature space 42
3.2.1. Neural classifiers 46
3.2.1.1. Multilayer neural classifiers 49
3.2.1.2. Teaching of a multi-layer network 51
3.2.2. The SVM 54 method
3.3. Probabilistic recognition methods 58
3.3.1. Non-parametric estimation of intra-class distributions 59
3.3.2. Parametrical estimation of intra-class distributions 61
4. Classification of vector sequences 64
4.1. Dynamic programming 61
4.2. Implicit Markov models (HMMs) 69
4.2.1. Sequence recognition 71
4.2.2. Model training 73
Part II Basic methods of biometry
5. Introduction 77
5.1. Biometrics characteristics 78
5.2. Biometric systems 79
5.2.1. Criteria for choosing biometric technologies 80
5.2.2. Criteria for assessing the correctness of biometric analysis 81
5.2.3. Comparison of properties of biometric technologies used 83
6. Biometrics using physical characteristics 85
6.1. Automatic fingerprint analysis 86
6.1.1. Methods of fingerprint acquisition 87
6.1.2. Fingerprint classification 89
6.1.3. Fingerprint recognition 89
6.1.3.1. Comparing the fingerprint 90
6.1.3.2. Comparison of the minutiae sets 91
6.2. Automatic recognition of the iris 96
6.2.1. Iris image acquisition 97
6.2.2. Iris recognition 99
6.2.2.1. Iris image segmentation 99
6.2.2.2. Designation of iris code 101
6.2.2.3. Iris recognition 104
6.3. Automatic face recognition 106
6.3.1. Structural facial modeling 110
6.3.2. Statistical and probabilistic facial modeling 111
6.3.2.1. Face recognition using the PCA 112 method
6.3.2.2. Face recognition systems 114
6.4. Recognition based on DNA analysis 116
6.4.1. Individual DNA profile 117
6.4.2. Determining the DNA profile 118
6.4.3. Perspectives of biometric applications of DNA 119
7. Biometrics using behavioral characteristics 121
7.1. Voice-based recognition 122
7.1.1. Methods for the quantitative description of the speech signal 124
7.1.2. Methodology of speaker recognition 127
7.1.2.1. Speaker recognition independent of the spoken text 128
7.1.2.2. Identifying the speaker on the basis of established statements 129
7.2. Recognition based on style analysis and dynamics of signing 130
7.2.1. Automatic analysis of the signature form 132
7.2.2. Automatic analysis of signature dynamics 133
Close 135
Literature 137