Computer Vision in Control Systems-3 (eBook)

Preface 6
Contents 7
About the Editors 12
1 Theoretical and Practical Solutions in Remote Sensing 14
Abstract 14
1.1 Introduction 14
1.2 Chapters Included in the Book 15
1.3 Conclusions 20
References 20
2 Multidimensional Image Models and Processing 23
Abstract 23
2.1 Introduction 24
2.2 Mathematical Models of Images 26
2.2.1 Random Fields 26
2.2.2 Tensor Models of Random Fields 27
2.2.3 Autoregressive Models of Random Fields 30
2.2.4 Wave Models of Random Fields 37
2.2.5 Random Fields on Surfaces 40
2.3 Image Filtering 44
2.3.1 Efficiency of Optimal Image Filtering 45
2.3.2 Tensor Kalman Filter 47
2.4 Anomalies Detection in Noisy Background 49
2.4.1 Optimal Algorithms for Signal Detection 50
2.4.2 Efficiency of Anomaly Detection 54
2.5 Image Alignment 57
2.5.1 Tensor Shift Filtering 57
2.5.2 Random Field Alignment of Images with Interframe Geometric Transformation 59
2.5.3 Alignment of Two Frames of Gaussian Random Field 60
2.5.4 Method of Fixed Point at Frame Alignment 63
2.6 Adaptive Algorithms of Image Processing 67
2.6.1 Pseudo-Gradient Adaptive Algorithms 68
2.6.2 Pseudo-Gradient Adaptive Prediction Algorithms 71
2.6.3 Pseudo-Gradient Algorithms of Image Alignment 73
2.7 Conclusions 74
Acknowledgements 75
References 75
3 Vision-Based Change Detection Using Comparative Morphology 77
Abstract 77
3.1 Introduction 78
3.2 Related Works 79
3.3 Methodology 81
3.3.1 Comparative Morphology as a Generalized Scheme of Morphological Image Analysis 82
3.3.2 Comparative Filtering as a Generalization of Morphological Filtering 84
3.3.3 Comparative Filters Based on Guided Contrasting 86
3.3.4 Manifold Learning, Diffusion Maps and Shape Matching 88
3.3.5 Generalized Diffusion Morphology and Comparative Filters Based on Diffusion Operators 91
3.3.6 Image and Shape Matching Based on Diffusion Morphology 94
3.3.7 Change Detection Pipeline Based on Comparative Filtering 96
3.4 Experiments 98
3.4.1 Qualitative Change Detection Experiments 98
3.4.2 Quantitative Change Detection Experiments 99
3.5 Conclusions 103
Acknowledgements 103
References 104
4 Methods of Filtering and Texture Segmentation of Multicomponent Images 109
Abstract 109
4.1 Introduction 110
4.2 Method for Nonlinear Multidimensional Filtering of Images 111
4.2.1 Mathematical Model of Multispectral Images 112
4.2.2 Nonlinear Multidimensional Filtering 114
4.3 Method of Texture Segmentation 122
4.4 Conclusions 128
References 129
5 Extraction and Selection of Objects in Digital Images by the Use of Straight Edges Segments 131
Abstract 131
5.1 Introduction 132
5.2 Related Works 133
5.3 Problem Statement and Method of Solution 136
5.3.1 Problem Statement and Tasks 136
5.3.2 Image Processing Structure 136
5.3.3 Advanced Algorithm of Straight Edge Segments Extraction 138
5.3.4 Lines Grouping Algorithms for Object Description and Selection 141
5.4 Modelling of Straight Edge Segments Extraction 143
5.5 Modelling of Segment Grouping and Object Detection, Selection and Localization 146
5.6 Experimental Results for Aerial, Satellite and Radar Images 149
5.7 Conclusions 155
Acknowledgements 157
References 157
6 Automated Decision Making in Road Traffic Monitoring by On-board Unmanned Aerial Vehicle System 160
Abstract 160
6.1 Introduction 161
6.2 Algorithm for Classification of Road Traffic Situation 163
6.3 Alphabet Forming for Description of Situation Class 164
6.4 Detection of Scene Objects 166
6.5 Forming Observed Scene Descriptions 170
6.6 Decision Making Procedure 174
6.7 Use of Statistical Criteria for Decision Making 175
6.8 Use of Functional Criteria Model 178
6.9 Examples 181
6.10 Conclusions 184
References 184
7 Warping Techniques in Video Stabilization 187
Abstract 187
7.1 Introduction 187
7.2 Development of Warping Techniques 189
7.3 Related Works 191
7.3.1 Overview of Warping Techniques in 2D Stabilization 191
7.3.2 Overview of Warping Techniques in 3D Stabilization 193
7.3.3 Overview of Inpainting Methods for Boundary Completion 197
7.4 Scene Classification 199
7.4.1 Extraction of Key Frames 199
7.4.2 Estimation of Scene Depth 201
7.5 3D Warping During Stabilization 202
7.5.1 Extraction of Structure from Motion 203
7.5.2 3D Camera Trajectory Building 205
7.5.3 Warping in Scenes with Small Depth 206
7.5.4 Warping Using Geometrical Primitives 207
7.5.5 Local 3D Warping in Scenes with Large Depth 210
7.6 Warping During Video Inpainting 213
7.7 Experimental Results 218
7.8 Conclusions 221
Acknowledgements 222
References 222
8 Image Deblurring Based on Physical Processes of Blur Impacts 226
Abstract 226
8.1 Introduction 227
8.2 Overview of Blur and Deblurring Methods 228
8.3 Model of Linear Blur Formation 237
8.3.1 Discrete-Analog Representation of Illumination 237
8.3.2 Small Linear Blur 239
8.3.3 Large Linear Blur 246
8.4 Non-linear Blur 250
8.5 Conclusions 257
References 257
9 Core Algorithm for Structural Verification of Keypoint Matches 260
Abstract 260
9.1 Introduction 261
9.2 Overview of Keypoint-Based Methods 262
9.3 Notation 264
9.4 Training Datasets 268
9.5 Algorithms of Primary Outlier Elimination 269
9.5.1 Nearest Neighbour Ratio Test 269
9.5.2 Crosscheck 270
9.5.3 Exclusion of Ambiguous Matches 270
9.5.4 Comparison and Conclusion 272
9.6 Geometrical Verification of Matches 273
9.6.1 Hough Clustering of Keypoint Matches 273
9.6.2 Verification of Cluster Hypotheses 278
9.7 Confidence Estimation 282
9.8 Application for Practical Tasks 284
9.8.1 Matching of Images of 3D Scenes 285
9.8.2 Image Retrieval Using Bag of Words 286
9.8.3 Registration of Aerospace Images 289
9.9 Conclusions 292
Acknowledgements 292
References 292
10 Strip-Invariants of Double-Sided Matrix Transformation of Images 296
Abstract 296
10.1 Introduction 297
10.2 Double-Sided Matrix Transformation of Images 298
10.3 Invariant Images (Problem Statement) 301
10.4 Analysis of Strip-Invariant Images 307
10.4.1 Existence Criteria of Strip-Invariant Images for Case m = n, A = B 307
10.4.2 Renunciation of Matrices Equality Condition A = B 316
10.4.3 Abandonment of Conditions of Symmetry and Orthogonality of Matrices A and B 319
10.5 Getting Invariant Images Using Eigenvectors of Matrices 321
10.6 Synthesis of Matrices by Given Invariant Images 325
10.7 Conclusions 327
References 327
11 The Fibonacci Numeral System for Computer Vision 329
Abstract 329
11.1 Introduction 330
11.2 Fibonacci Numbers 333
11.3 Fibonacci Numbers and Fibonacci Codes 334
11.4 Minimal and Maximum Representation Forms for Fibonacci Numbers 336
11.4.1 Estimate of the Noise Immunity of the Fibonacci Codes in Their Minimal Form 337
11.4.2 Noise-Proof Fibonacci Counting 343
11.5 Decoding the Fibonacci Combinations in the Minimal Representation Form 346
11.6 Conclusions and Future Research 350
Acknowledgements 351
References 351