Table of contents for image processing, analysis and machine vision

Chapter 1 Introduction 1

1.1 Motivation 1

1.2 Why computer vision is difficult 2

1.3 Image expression and image analysis Task 4

1.4 Summary 7

1.5 Reference 7

Chapter 2 Images, Their Expression and Properties 8

2.1 Images Express several concepts 8

Continuous image function 8

2.2 Image digitization 10

2.2.1 Sampling 10

2.2.2 Quantization 11

2.3 Digital image properties 12

2.3.1 Metric sum of digital images

Topological properties 12

2.3.2 Histogram 16

2.3.3 Entropy17

2.3.4 Visual perception of images18

2.3.5 Image quality20

2.3.6 Noise in images 20

2.4 Color images 22

2.4.1 Color physics 22

2.4.2 Colors perceived by humans 23

2.4.3 Color space 26

2.4.4 Palette image 28

2.4.5 Color constancy 28

2.5 Camera overview 29

2.5.1 Photosensitive sensor 29

2.5.2 Black and white camera 30

2.5.3 Color camera 32

2.6 Summary 33

2.7 References 34

Chapter 3 Images and their mathematical and physical background 35

3.1 Overview 35

3.1.1 Linear 35

3.1.2 Dirac distribution and convolution 35

3.2 Integral linear transformation 37

3.2.1 As linear Image of the system 37

3.2.2 Introduction to integral linear transformation 37

3.2.3 1D Fourier transform 38

3.2.4 2 D Fourier transform Transformation 41

3.2.5 Sampling and Shannon constraints 43

3.2.6 Discrete cosine transform 46

3.2.7 Wavelet transform 47

3.2.8 Eigen analysis 51

3.2.9 Singular value decomposition 52

3.2.10 Principal component analysis 53

3.2.11 Other orthogonal images Transformation 54

3.3 Image as stochastic process 55

3.4 Physics of image formation 57

3.4.1 Image as radiation measurement 57

3.4.2 Image acquisition and geometric optics 57

3.4.3 Lens aberration and radial distortion 60

3.4.4 Image acquisition from a radiological perspective 62

3.4.5 Surface Reflection 64

3.5 Summary 67

3.6 References 67

Chapter 4 Data Structure of Image Analysis 69

4.1 Levels of image data representation 69

4.2 Traditional image data structure 70

4.2.1 Matrix 70

4.2.2 Chain 72

4.2.3 Topological data structure 73

4.2.4 Relational structure 73

4.3 Hierarchical data structure 74

4.3.1 Pyramid 74

4.3.

2 Quadtree75

4.3.3 Other pyramid structures76

4.4 Summary77

4.5 References78

Chapter 5 Image preprocessing 79

5.1 Pixel brightness transformation 79

5.1.1 Position-related brightness correction 80

5.1.2 Grayscale transformation 80

5.2 Geometric transformation 82

5.2.1 Pixel coordinate transformation 83

5.2.2 Brightness interpolation 84

5.3 Local preprocessing 86

5.3.1 Image smoothing 86

5.3.2 Edge detection operator 92

5.3.3 Second-order derivative zero-crossing point 96

5.3. 4 Scale in image processing 98

5.3.5 Canny edge extraction 100

5.3.6 Parametric edge model 102

5.3.7 In multispectral images The edge of >

5.3.10 Corner point (interest point) detection 109

5.3.11 Maximum stable extreme value area detection 112

5.4 Image restoration 114

5.4.1 Easily recoverable degradation 114

5.4.2 Inverse filtering 115

5.4.3 Wiener filtering 115

5.5 Summary 117

5.6 References 118

Chapter 6 Segmentation I 124

6.1 Thresholding 124

6.1.1 Threshold detection method 126

6.1.2 Optimal thresholding127

6.1.3 Multispectral thresholding129

6.2 Edge-based segmentation130

6.2.1 Edge image threshold 131

6.2.2 Edge relaxation method 133

6.2.3 Boundary tracking 135

6.2.4 Edge tracking as graph search 139

6.2.5 Edge tracking as dynamic programming 146

6.2.6 Hough transform 149

6.2.7 Boundary detection using boundary position information

Boundary detection 155

6.2.8 Constructing regions from boundaries 156

6.3 Region-based segmentation 157

6.3.1 Region merging 158

6.3 .2 Region split 160

6.3.3 Split and merge 161

6.3.4 Watershed segmentation 163

6.3.5 Region growth post-processing 166

6.4 Matching 166

6.4.1 Matching criteria 167

6.4.2 Matching control strategy 168

6.5 Segmentation evaluation questions 169

6.5.1 Supervised evaluation 169

6.5.2 Unsupervised evaluation 172

6.6 Summary 172

6.7 References 175

Chapter 7 Segmentation II 182

7.1 Mean Shift Segmentation 182

7.2 Active Contour Model - Snake 187

7.2.1 Classic Snakes and Balloons 188

7.2.2 Extensions 191

7.2.3 Gradient Vector Flow Snakes 191

7

.3 Geometric deformation model - level set sum

Geodesic active contour 194

7.4 Fuzzy connectivity 200

7.5 Towards image segmentation based on 3D graphs 204

7.5.1 Simultaneous detection of boundary pairs 205

7.5.2 Suboptimal surface detection 208

7.6 Graph cut segmentation 209

7.7 Optimal single and multi-surface segmentation 214

7.8 Summary 223

7.9 References 224

Chapter 8 Shape Representation and Description 232

8.1 Region identification 234

8.2 Contour-based shape representation and description 236

8.2.1 Chain code 237

8.2.2 Simple geometric boundary representation 237

8.2.3 Fourier transform of boundaries 239

8.2.4 Boundary description using fragment sequences 241

8.2.5 B-spline representation 243

8.2.6 Other contour-based shapes

Description methods 245

8.2.7 Shape invariants 245

8.3 Region-based Shape representation and description 248

8.3.1 Simple scalar region description 248

8.3.2 Moment 251

8.3.3 Convex hull 253

8.3.4 Graph representation based on region skeleton 257

8.3.5 Region decomposition 259

8.3.6 Region neighborhood graph 260

8.4 Shape category 261

8.5 Summary 261

8.6 References 263

Chapter 9 Object Recognition 270

9.1 Knowledge Representation 270

9.2 Statistical pattern recognition 274

9.2.1 Classification principle 275

9.2.2 Classifier settings 276

9.2.3 Classifier learning 278

9.2.4 Support vector machine 280

9.2.5 Cluster analysis 284

9.3 Neuron network 286

9.3.1 Previous Feed network 287

9.3.2 Unsupervised learning 288

9.3.3 Hopfield neuron network 289

9.4 Syntactic pattern recognition 290

9.4.1 Grammar and Language 291

9.4.2 Syntactic Analysis and Syntactic Classifier 293

9.4.3 Syntactic Classifier Learning and

Grammar Derivation 294

9.5 Recognition as graph matching 295

9.5.1 Isomorphism of graphs and subgraphs 296

9.5.2 Similarity of graphs 298

9.6 Optimization technology in recognition 299

9.6.1 Genetic algorithm 300

9.6.2 Simulated annealing 302

9.7 Fuzzy system 303

9.7.1 Fuzzy sets and fuzzy membership functions 304

9.7.2 Fuzzy set operations 305

9.7.3 Fuzzy reasoning 306

9.7. 4 Fuzzy system design and training 308

9.8 Boosting method in pattern recognition 309

9.9 Summary 311

9.10 References 314

Chapter 10 Image Understanding 319

10.1 Image Understanding Control Strategy 320

<

p>10.1.1 Parallel and serial processing control 320

10.1.2 Hierarchical control 321

10.1.3 Bottom-up control 321

10.1 .4 Model-based control 321

10.1.5 Hybrid control strategies 322

10.1.6 Non-stratified control 325

10.2 RANSAC: via random sampling Consistent

to fit326

10.3 Point distribution model329

10.4 Activity appearance model337

10.5 Patterns in image understanding Recognition methods 344

10.5.1 Classification-based segmentation 344

10.5.2 Contextual image classification 346

10.6 Boosted cascade classifier for fast

Object detection 349

10.7 Scene labeling and constraint propagation 352

10.7.1 Discrete relaxation method 353

10.7.2 Probabilistic relaxation method 355

p>

10.7.3 Search interpretation tree 357

10.8 Semantic image segmentation and understanding 357

10.8.1 Semantic region growth 358

10.8.2 Genetic image interpretation 360

10.9 Hidden Markov model 365

10.9.1 Application 369

10.9.2 Coupled HMM 370

10.9.3 Bayesian Belief Network 371

10.10 Gaussian Mixture Model and Expectation Maximization 372

10.11 Summary 378

10.12 References 380

Chapter 11 3D Vision and Geometry 389

11.1 3D Vision Tasks 389

11.1.1 Marr Theory 391

11.1.2 Other Vision Categories : Active and

Purposeful vision 392

11.2 Fundamentals of projective geometry 393

11.2.1 Points and hyperplanes in projective space 394

11.2.2 Homography 395

11.2.3 Estimating homography from corresponding points 397

11.3 Single perspective camera 400

11.3. 1 Camera model 400

11.3.2 Projection and back projection in homogeneous coordinate system 402

11.3.3 Calibrating a scene from a known scene

p>

Camera 403

11.4 Scene reconstruction from multiple views 403

11.4.1 Triangulation 403

11.4.2 Projective reconstruction 404

11.4.3 Matching constraints 405

11.4.4 Beam adjustment method 406

11.4.5 Upgraded projective reconstruction and

Self-calibration 407

11.5 Dual Cameras and Stereo Perception 408

11.5.1 Epipolar Geometry - Basics

Matrix 408

11.5.2 Camera Relative motion of Corresponding point estimation basic matrix 411

11.5.5 Dual-camera correction structure 412

11.5.6 Correction calculation 414

11.6 Three-camera and three-view tensor 415

1

1.6.1 Stereo corresponding point algorithm 417

11.6.2 Active acquisition of range images 421

11.7 From radiometric measurement to 3D information 423

11.7.1 From Shadow to Shape 423

11.7.2 Photometric Stereovision 426

11.8 Summary 427

11.9 References 428

Chapter 12 Application of 3D Vision 433

12.1 From X to Shape 433

12.1.1 From Motion to Shape 433

12.1.2 From Texture to Shape 437

p>

12.1.3 Other X-to-shape technologies 439

12.2 Complete 3D objects 440

12.2.1 3D objects and models And

Related issues 440

12.2.2 Line annotation 441

12.2.3 Volume representation and direct measurement 443

12.2.4 Volume modeling strategy 444

12.2.5 Surface modeling strategy 446

12.2.6 To obtain a complete 3D model

Bell labeling and fusion 447

12.3 Vision based on 3D model 451

12.3.1 General considerations 451

12.3.2 Goad algorithm 452

12.3.3 Based on Luminance image of the model

Curved surface object recognition 455

12.3.4 Model-based distance

Image recognition 456

12.4 3D scene 2D view representation 456

12.4.1 Observation space 456

12.4.2 Multi-view representation and representation 457

12.4.3 Structure as 2D view

Expressed geometric primitives 457

12.4.4 Using stored 2D views to display 3D real-world scenes 458

12.5 Examples Research - 3D Reconstruction from Unorganized 2D View Sets 460

12.6 Summary 463

12.7 References 464

Chapter 13 Chapter Mathematical Morphology 470

13.1 Basic Concepts of Morphology 470

13.2 Four Principles of Morphology 471

13.3 Binary Expansion and Corrosion 472

13.3.1 Expansion 472

13.3.2 Corrosion 474

13.3.3 Hit and miss transformation 476

13.3.4 Opening and closing operations Operation 476

13.4 Gray scale expansion and corrosion 477

13.4.1 Top surface, umbra, gray scale

Dilation and corrosion 477

13.4.2 The umbral homeomorphism theorem and properties of expansion,

corrosion and opening and closing operations

479

13.4.3 Top hat Transformation 480

13.5 Skeleton and object labeling 481

13.5.1 Homotopy transformation 481

13.5.2 Skeleton and maximum sphere 481

13.5.3 Refinement, coarsening and homotopy skeleton 482

13.5.4 Extinction function and final corrosion 485

13.5.5 Final corrosion and distance function 486

13.5.6 Geodesic transformation 487

13.5.7 Morphological reconstruction 488

13.6 grains

Measuring method 489

13.7 Morphological segmentation and watershed 491

13.7.1 Particle segmentation, labeling and watershed 491

13.7 .2 Binary morphological segmentation 491

13.7.3 Gray-level segmentation and watershed 493

13.8 Summary 494

13.9 References 495

Chapter 14 Image Data Compression 497

14.1 Properties of Image Data 498

14.2 Discretization in Image Data Compression

Image Transformation 498

14.3 Predictive compression methods 500

14.4 Vector quantization 502

14.5 Hierarchical and progressive compression methods 502

14.6 Comparison of compression methods 503

p>

14.7 Other technologies 504

14.8 Encoding 504

14.9 JPEG and MPEG image compression 505

14.9.1 JPEG - still images

p>

Compression 505

14.9.2 JPEG-2000 Compression 506

14.9.3 MPEG - Full Motion

Video Compression 508

p>

14.10 Summary 509

14.11 References 511

Chapter 15 Texture 514

15.1 Statistical texture description 516

15.1.1 Method based on spatial frequency 516

15.1.2 Generated matrix 517

15.1.3 Edge frequency 519

15.1.4 Primitive element Length (stroke) 520

15.1.5 Laws texture energy measurement 521

15.1.6 Fractal texture description 521

15.1.7 Multi-scale texture description

p>

——Wavelet domain method 522

15.1.8 Other texture description methods

Statistical methods 525

15.2 Syntactic texture description method 526

p>

15.2.1 Shape chain syntax 526

15.2.2 Graph syntax 527

15.2.3

Primitive grouping in layered textures 528

15.3 Mixed texture description method 530

15.4 Application of texture identification method 531

15.5 Summary 531

15.6 References 532

Chapter 16 Motion Analysis 537

16.1 Differential Motion Analysis Method 539

16.2 Optical Flow 542

16.2.1 Optical Flow Calculation 542

16.2.2 Global and local optical flow estimation544

16.2.3 Combined local and global optical flow estimation546

16.2.4 Optical flow in motion analysis546

16.3 Analysis based on correspondence between interest points549

16.3.1 Detection of interest points549

16.3.2 Correspondence of points of interest 549

16.4 Detection of specific motion patterns 551

16.5 Video tracking 554

16.5.1 Background modeling 554

p>

16.5.2 Tracking based on kernel function 558

16.5.3 Target path analysis 562

16.6 Motion model to assist tracking 566

<

p>16.6.1 Kalman filter 567

16.6.2 Particle filter 570

16.7 Summary 573

16.8 References 575

Vocabulary 581