Digital Image Processing, 4th edition

Gonzalez and Woods

Pearson/Prentice Hall
© 2018

Table of Contents
Chapter 1 Introduction 1
1.1 What is Digital Image Processing? 2
1.2 The Origins of Digital Image Processing 3
1.3 Examples of Fields that Use Digital Image Processing 7
Gamma-Ray Imaging 8
X-Ray Imaging 8
Imaging in the Ultraviolet Band 11
Imaging in the Visible and Infrared Bands 12
Imaging in the Microwave Band 17
Imaging in the Radio Band 18
Other Imaging Modalities 19
1.4 Fundamental Steps in Digital Image Processing 25
1.5 Components of an Image Processing System 28

Chapter 2 Digital Image Fundamentals 31

2.1 Elements of Visual Perception 32
Structure of the Human Eye 32
Image Formation in the Eye 34
Brightness Adaptation and Discrimination 34
2.2 Light and the Electromagnetic Spectrum 38
2.3 Image Sensing and Acquisition 41
Image Acquisition Using a Single Sensing Element 42
Image Acquisition Using Sensor Strips 44
Image Acquisition Using Sensor Arrays 45
A Simple Image Formation Model 45
2.4 Image Sampling and Quantization 47
Basic Concepts in Sampling and Quantization 47
Representing Digital Images 49
Linear vs. Coordinate Indexing 54
Spatial and Intensity Resolution 55
Image Interpolation 61
2.5 Some Basic Relationships Between Pixels 63
 Neighbors of a Pixel 63
Adjacency, Connectivity, Regions, and Boundaries 63
Distance Measures 66
2.6 Introduction to the Basic Mathematical Tools Used in Digital Image Processing 67
Elementwise versus Matrix Operations 67
Linear versus Nonlinear Operations 68
Arithmetic Operations 69
Set and Logical Operations 75
Basic Set Operations 75
Logical Operations 80
Fuzzy Sets 82
Spatial Operations 83
Single-Pixel Operations 83
Neighborhood Operations 83
Geometric Transformations 84
Image Registration 88
Vector and Matrix Operations 90
Image Transforms 93
Probability and Random Variables 96
Sample Spaces, Events, and Probability 96
The Sum (Addition) Rule of Probability 97
Conditional Probability 98
Independence 100
The Law of Total Probability 101
Bayes’ Rule 102
Random Variables 103
Probability Functions for Discrete Random Variables 105
Some Important Probability Mass Functions 105
Estimating Discrete Probability Functions from Sample Data 106
Expected Value and Moments of Discrete Random Variables 107
Continuous Random Variables 110
The Uniform and Gaussian Probability Density Functions 111
Expected Values and Moments of Continuous Random Variables 114
Estimating the Mean, Variance, and Higher-Order Moments from Sample Data 115
Multivariate Random Variables 117
The Multivariate Gaussian PDF 118
Estimating the Parameters of the Multivariate Gaussian PDF 120

Chapter 3 Intensity Transformations and Spatial Filtering

3.1 Background 134
The Basics of Intensity Transformations and Spatial Filtering 134
About the Examples in this Chapter 136
3.2 Some Basic Intensity Transformation Functions 136
 Image Negatives 136
Log Transformations 138
Power-Law (Gamma) Transformations 139
Piecewise Linear Transformation Functions 142
Contrast Stretching 143
Intensity-Level Slicing 144
Bit-Plane Slicing 145
3.3 Histogram Processing 147
Histogram Equalization 148
Histogram Matching (Specification) 156
Exact Histogram Matching (Specification) 163
Foundation 165
Ordering 165
Computing the neighborhood averages and extracting the K-tuples: 167
Exact Histogram Specification Algorithm 168
Local Histogram Processing 173
Using Histogram Statistics for Image Enhancement 174
3.4 Fundamentals of Spatial Filtering 177
The Mechanics of Linear Spatial Filtering 178
Spatial Correlation and Convolution 178
Separable Filter Kernels 185
Some Important Comparisons Between Filtering in the Spatial and Frequency Domains 186
A Word about how Spatial Filter Kernels are Constructed 188
3.5 Smoothing (Lowpass) Spatial Filters 188
Box Filter Kernels 189
Lowpass Gaussian Filter Kernels 190
Order-Statistic (Nonlinear) Filters 198
3.6 Sharpening (Highpass) Spatial Filters 199
Foundation 200
Using the Second Derivative for Image Sharpening—The Laplacian 202
Unsharp Masking and Highboost Filtering 206
Using First-Order Derivatives for Image Sharpening—The Gradient 208
3.7 Highpass, Bandreject, and Bandpass Filters from Lowpass Filters 212
3.8 Combining Spatial Enhancement Methods 216
3.9 Using Fuzzy Techniques for Intensity Transformations and Spatial Filtering
Introduction 220
Principles of Fuzzy Set Theory 221
Definitions 221
Some Common Membership Functions 223
Using Fuzzy Sets 224
Using Fuzzy Sets for Intensity Transformations 233
Using Fuzzy Sets for Spatial Filtering 236
Chapter 4 Filtering in the Frequency Domain
4.1 Background 250
A Brief History of the Fourier Series and Transform 250
About the Examples in this Chapter 252
4.2 Preliminary Concepts 253
Complex Numbers 253
Fourier Series 254
Impulses and their Sifting Properties 254
The Fourier Transform of Functions of One Continuous Variable 256
Convolution 259
4.3 Sampling and the Fourier Transform of Sampled Functions 261
Sampling 261
The Fourier Transform of Sampled Functions 262
The Sampling Theorem 263 Aliasing 267
Function Reconstruction (Recovery) from Sampled Data 270
4.4 The Discrete Fourier Transform of One Variable 271
Obtaining the DFT from the Continuous Transform of a Sampled Function 271
Relationship Between the Sampling and Frequency Intervals 274
4.5 Extensions to Functions of Two Variables 276
The 2-D Impulse and Its Sifting Property 276
The 2-D Continuous Fourier Transform Pair 277
2-D Sampling and the 2-D Sampling Theorem 277
Aliasing in Images 279
Extensions from 1-D Aliasing 279
Image Resampling and Interpolation 283
Aliasing and Moiré Patterns 284
The 2-D Discrete Fourier Transform and Its Inverse 286
4.6 Some Properties of the 2-D DFT and IDFT 286
Relationships Between Spatial and Frequency Intervals 286
Translation and Rotation 287
Periodicity 287
Symmetry Properties 289
Fourier Spectrum and Phase Angle 295
The 2-D Discrete Convolution Theorem 299
Summary of 2-D Discrete Fourier Transform Properties 303
4.7 The Basics of Filtering in the Frequency Domain 306
Additional Characteristics of the Frequency Domain 306
Frequency Domain Filtering Fundamentals 307
Summary of Steps for Filtering in the Frequency Domain 312
Correspondence Between Filtering in the Spatial and
Frequency Domains 314
 4.8 Image Smoothing Using Lowpass Frequency Domain Filters 318
Ideal Lowpass Filters 319
Gaussian Lowpass Filters 323
Butterworth Lowpass Filters 324
Additional Examples of Lowpass Filtering 327
4.9 Image Sharpening Using Highpass Filters 330
Ideal, Gaussian, and Butterworth Highpass Filters from Lowpass Filters 330
The Laplacian in the Frequency Domain 335
Unsharp Masking, High-boost Filtering, and High-Frequency-Emphasis Filtering 337
Homomorphic Filtering 339
4.10 Selective Filtering 342
Bandreject and Bandpass Filters 343
Notch Filters 345
4.11 The Fast Fourier Transform 349
Separability of the 2-D DFT 349
Computing the IDFT Using a DFT Algorithm 350
The Fast Fourier Transform (FFT) 350

Chapter 5 Image Restoration and Reconstruction

5.1 A Model of the Image Degradation/Restoration Process 366
5.2 Noise Models 366
Spatial and Frequency Properties of Noise 367
Some Important Noise Probability Density Functions 367
Gaussian Noise 367
Rayleigh Noise 368
Erlang (Gamma) Noise 369
Exponential Noise 369
Uniform Noise 369
Salt-and-Pepper Noise 370
Periodic Noise 372
Estimating Noise Parameters 373
5.3 Restoration in the Presence of Noise Only------Spatial Filtering 375
Mean Filters 376
Arithmetic Mean Filter 376
Geometric Mean Filter 376
Harmonic Mean Filter 377
Contraharmonic Mean Filter 377
Order-Statistic Filters 378
Median Filter 378
Max and Min Filters 380
Midpoint Filter 380
Alpha-Trimmed Mean Filter 380
 Adaptive Filters 381
Adaptive, Local Noise Reduction Filter 384
Adaptive Median Filter 386
5.4 Periodic Noise Reduction Using Frequency Domain Filtering 388
More on Notch Filtering 389
Optimum Notch Filtering 393
5.5 Linear, Position-Invariant Degradations 396
5.6 Estimating the Degradation Function 400
Estimation by Image Observation 400
Estimation by Experimentation 400
Estimation by Modeling 401
5.7 Inverse Filtering 404
5.8 Minimum Mean Square Error (Wiener) Filtering 406
5.9 Constrained Least Squares Filtering 411
5.10 Geometric Mean Filter 415
5.11 Image Reconstruction from Projections 416
Introduction 416
Principles of X-ray Computed Tomography (CT) 418
Projections and the Radon Transform 422
Backprojections 425
The Fourier-Slice Theorem 427
Reconstruction Using Parallel-Beam Filtered Backprojections 428
Reconstruction Using Fan-Beam Filtered Backprojections 434

Chapter 6 Wavelet and Other Image Transforms

6.1 Preliminaries 452
6.2 Matrix-based Transforms 454
Rectangular Arrays 460
Complex Orthonormal Basis Vectors 461
Biorthonormal Basis Vectors 462
6.3 Correlation 466
6.4 Basis Functions in the Time-Frequency Plane 467
6.5 Basis Images 471
6.6 Fourier-Related Transforms 472
The Discrete hartley Transform 473
The Discrete Cosine Transform 475
The Discrete Sine Transform 480
6.7 Walsh-Hadamard Transforms 484
6.8 Slant Transform 488
 6.9 Haar Transform 490
6.10 Wavelet Transforms 492
Scaling Functions 493
Wavelet Functions 495
Wavelet Series Expansion 498
Discrete Wavelet Transform in One Dimension 500
The Fast Wavelet Transform 501
Wavelet Transforms in Two Dimensions 508
Wavelet Packets 514

Chapter 7 Color Image Processing 529

7.1 Color Fundamentals 530
7.2 Color Models 535
The RGB Color Model 537
The CMY and CMYK Color Models 538
The HSI Color Model 541
Converting Colors from RGB to HSI 543
Converting Colors from HSI to RGB 545
Manipulating HSI Component Images 547
A Device Independent Color Model 548
7.3 Pseudocolor Image Processing 550
Intensity Slicing and Color Coding 550
Intensity to Color Transformations 553
7.4 Basics of Full-Color Image Processing 559
7.5 Color Transformations 560
Formulation 560
Color Complements 564
Color Slicing 566
Tone and Color Corrections 567
Histogram Processing of Color Images 569
7.6 Color Image Smoothing and Sharpening 572
Color Image Smoothing 572
Color Image Sharpening 574
7.7 Using Color in Image Segmentation 575
Segmentation in HSI Color Space 576
Segmentation in RGB Space 576
Color Edge Detection 580
7.8 Noise in Color Images 582
7.9 Color Image Compression 585

Chapter 8 Image Compression and Watermarking

8.1 Fundamentals 596
Coding Redundancy 597
Spatial and Temporal Redundancy 599
Irrelevant Information 600
Measuring Image Information 601
Shannon’s First Theorem 602
Fidelity Criteria 603
Image Compression Models 605
The Encoding or Compression Process 606
The Decoding or Decompression Process 607
Image Formats, Containers, and Compression Standards 607
8.2 Huffman Coding 609
8.3 Golomb Coding 612
8.4 Arithmetic Coding 617
Adaptive context dependent probability estimates 618
8.5 LZW Coding 620
8.6 Run-length Coding 622
One-dimensional CCITT compression 624
Two-dimensional CCITT compression 624
8.7 Symbol-based Coding 628
JBIG2 compression 629
8.8 Bit-plane Coding 631
8.9 Block Transform Coding 632
Transform selection 633
Subimage size selection 638
Bit allocation 639
Zonal Coding Implementation 640
Threshold Coding Implementation 641
JPEG 644
8.10 Predictive Coding 650
Lossless predictive coding 650
Motion compensated prediction residuals 655
Lossy predictive coding 661
Optimal predictors 665
Optimal quantization 667
8.11 Wavelet Coding 670
Wavlet selection 671
Decomposition level selection 672
Quantizer design 673
JPEG-2000 674
8.12 Digital Image Watermarking 680
Chapter 9 Morphological Image Processing
9.1 Preliminaries 694
9.2 Erosion and Dilation 696
Erosion 697
Dilation 699
Duality 702
9.3 Opening and Closing 702
9.4 The Hit-or-Miss Transform 706
9.5 Some Basic Morphological Algorithms 710
Boundary Extraction 711
Hole Filling 711
Extraction of Connected Components 713
Convex Hull 715
Thinning 718
Thickening 718
Skeletons 720
Pruning 722
9.6 Morphological Reconstruction 725
Geodesic Dilation and Erosion 725
Morphological Reconstruction by Dilation and by Erosion 726
Sample Applications 727
Opening by Reconstruction 728
Automatic Algorithm for Filling Holes 729
Border Clearing 730
9.7 Summary of Morphological Operations on Binary Images 731
9.8 Grayscale Morphology 732
Grayscale Erosion and Dilation 732
Grayscale Opening and Closing 738
Some Basic Grayscale Morphological Algorithms 740
Morphological Smoothing 740
Morphological Gradient 740
Top-Hat and Bottom-Hat Transformations 741
Granulometry 743
Textural Segmentation 745
Grayscale Morphological Reconstruction 746

Chapter 10 Image Segmentation I: Edge Detection,

Thresholding, and Region Detection
10.1 Fundamentals 762
10.2 Point, Line, and Edge Detection 763
 Background 764
Detection of Isolated Points 768
Line Detection 769
Edge Models 772
Basic Edge Detection 778
The Image Gradient and Its Properties 778
Gradient Operators 779
Combining the Gradient with Thresholding 784
More Advanced Techniques for Edge Detection 786
The Marr-Hildreth Edge Detector 786
The Canny Edge Detector 791
Linking Edge Points 797
Local Processing 797
Global Processing Using the Hough Transform 799
10.3 Thresholding 804
Foundation 804
The Basics of Intensity Thresholding 805
The Role of Noise in Image Thresholding 806
The Role of Illumination and Reflectance in Image Thresholding 807
Basic Global Thresholding 808
Optimum Global Thresholding Using Otsu’s Method 809
Using Image Smoothing to Improve Global Thresholding 814
Using Edges to Improve Global Thresholding 815
Multiple Thresholds 819
Variable Thresholding 823
Variable Thresholding Based on Local Image Properties 823
Variable Thresholding Based on Moving Averages 825
10.4 Segmentation by Region Growing and by Region Splitting and Merging 826
Region Growing 826
Region Splitting and Merging 830
10.5 Region Segmentation Using Clustering and Superpixels 832
Region Segmentation using K-Means Clustering 832
Region Segmentation using Superpixels 834
SLIC Superpixel Algorithm 837
Specifying the Distance Measure 838
10.6 Region Segmentation Using Graph Cuts 840
Images as Graphs 841
Minimum Graph Cuts 844
Computing Minimal Graph Cuts 846
Graph Cut Segmentation Algorithm 847
10.7 Segmentation Using Morphological Watersheds 849
Background 850
 Dam Construction 853
Watershed Segmentation Algorithm 855
The Use of Markers 856
10.8 The Use of Motion in Segmentation 859
Spatial Techniques 859
A Basic Approach 859
Accumulative Differences 860
Establishing a Reference Image 861
Frequency Domain Techniques 862

Chapter 11 Image Segmentation II: Active Contours: Snakes

and Level Sets
11.1 Background 878
11.2 Image Segmentation Using Snakes 878
Explicit (Parametric) Representation of Active Contours 878
Derivation of the Fundamental Snake Equation 880
Iterative Solution of the Snake Equation 883
External Force Based on the Magnitude of the Image
Gradient (MOG) 888
External Force Based on Gradient Vector Flow (GVF) 896
11.3 Segmentation Using Level Sets 902
Implicit Representation of Active Contours 902
Derivation of the Level Set Equation 904
Discrete (Iterative) Solution of The Level Set Equation 908
Curvature 912
Specifying, Initializing, and Reinitializing Level Set Functions 913
Force Functions Based Only on Image Properties 917
Edge/Curvature-Based Forces 922
Region/Curvature-Based Forces 927
Improving the Computational Performance of Level Set Algorithms 938

Chapter 12 Feature Extraction

12.1 Background 954
12.2 Boundary Preprocessing 956
Boundary Following (Tracing) 956
Chain Codes 958
Freeman Chain Codes 958
Slope Chain Codes 961
Boundary Approximations Using Minimum-Perimeter Polygons 963
Foundation 963
MPP Algorithm 964
 Signatures 968
Skeletons, Medial Axes, and Distance Transforms 970
12.3 Boundary Feature Descriptors 973
Some Basic Boundary Descriptors 974
Shape Numbers 976
Fourier Descriptors 977
Statistical Moments 981
12.4 Region Feature Descriptors 982
Some Basic Descriptors 982
Topological Descriptors 985
Texture 988
Statistical Approaches 988
Spectral Approaches 997
Moment Invariants 1000
12.5 Principal Components as Feature Descriptors 1001
12.6 Whole-Image Features 1010
The Harris-Stephens Corner Detector 1011
Maximally Stable Extremal Regions (MSERs) 1018
12.7 Scale-Invariant Feature Transform (SIFT) 1023
Scale Space 1025
Detecting Local Extrema 1027
Finding the Initial Keypoints 1027
Improving the Accuracy of Keypoint Locations 1029
Eliminating Edge Responses 1031
Keypoint Orientation 1032
Keypoint Descriptors 1034
Summary of the SIFT Algorithm 1036

Chapter 13 Image Pattern Classification

13.1 Background 1050
13.2 Patterns and Pattern Classes 1052
Pattern Vectors 1052
Structural Patterns 1054
13.3 Pattern Classification by Prototype Matching 1056
Minimum-Distance Classifier 1056
Using Correlation for 2-D prototype matching 1061
Matching SIFT Features 1063
Matching Structural Prototypes 1065
Matching Shape Numbers 1065
String Matching 1066
13.4 Optimum (Bayes) Statistical Classifiers 1069
 Derivation of the Bayes Classifier 1069
Bayes Classifier for Gaussian Pattern Classes 1071
13.5 Neural Networks and Deep Learning 1077
Background 1077
The Perceptron 1080
Multilayer Feedforward Neural Networks 1089
Model of an Artificial Neuron 1089
Interconnecting Neurons to Form a Fully Connected Neural Network 1091
Forward Pass Through a Feedforward Neural Network 1094
The Equations of a Forward Pass 1094
Matrix Formulation 1096
Using Backpropagation to Train Deep Neural Networks 1099
The Equations of Backpropagation 1099
Matrix Formulation 1102
13.6 Deep Convolutional Neural Networks 1110
A Basic CNN Architecture 1110
Basics of How a CNN Operates 1111
Neural Computations in a CNN 1117
Multiple Input Images 1119
The Equations of a Forward Pass Through a CNN 1119
The Equations of Backpropagation Used to Train CNNs 1120
13.7 Some Additional Details of Implementation 1133

Bibliography 1143
Index 1157