A Selection of Image Processing Techniques: From Fundamental to Research Front

Cover
Half Title
Title Page
Copyright Page
Table of Contents
Preface
Author
Chapter 1◾ Introduction
	1.1 IMAGE BASICS
		1.1.1 Image Representation and Display
			1.1.1.1 Images and Pixels
			1.1.1.2 Matrix and Vector Representation of Image
			1.1.1.3 How the Image Is Displayed
		1.1.2 Spatial Resolution and Amplitude Resolution
			1.1.2.1 Sampling and Quantization
			1.1.2.2 Resolution and Data Volume
		1.1.3 Resolution and Image Quality
		1.1.4 Half-Tone and Dithering Technology
			1.1.4.1 Half-Tone Output Technology
			1.1.4.2 Half-T one Output Mask
			1.1.4.3 Dithering Technology
	1.2 IMAGE TECHNOLOGY
		1.2.1 Image Engineering
		1.2.2 Classification of Image Technology
		1.2.3 IP System
	1.3 CHARACTERISTICS OF THIS BOOK
		1.3.1 Writing Motivation
		1.3.2 Material Selection and Contents
		1.3.3 Structure and Arrangement
	REFERENCES
Chapter 2◾Image De-Noising
	2.1 NOISE TYPES AND CHARACTERISTICS
		2.1.1 Different Noises
		2.1.2 Noise Characteristics and Description
			2.1.2.1 Gaussian Noise
			2.1.2.2 Impulse (Salt and Pepper) Noise
			2.1.2.3 Uniform Noise
			2.1.2.4 Rayleigh Noise
	2.2 IMAGE FILTERING AND DE-NOISING
		2.2.1 Spatial Noise Filter
			2.2.1.1 Mean Filter
			2.2.1.2 Order Statistical Filter
			2.2.1.3 Hybrid Filter
			2.2.1.4 Mode Filter
		2.2.2 Frequency Domain Periodic Noise Filter
			2.2.2.1 Band-Pass Filter
			2.2.2.2 Band-Stop Filter
			2.2.2.3 Notch Filter
			2.2.2.4 Interactive Filtering
	2.3 SELECTIVE FILTER
	2.4 SWITCHING MEDIAN FILTER
		2.4.1 The Principle of Switching Median Filter
			2.4.1.1 Noise Model
			2.4.1.2 Noise Detection
			2.4.1.3 Noise Adaptive Filtering
		2.4.2 Switch-Based Adaptive Weighted Mean Filter
			2.4.2.1 Directional Differential Noise Detector
			2.4.2.2 Adaptive Weighted Mean Filter
		2.4.3 Further Improvements
			2.4.3.1 New Classification Criteria
			2.4.3.2 New Windows
			2.4.3.3 New Decision-Making Rules
			2.4.3.4 Comparison of Detection Results
	2.5 SOME RECENT DEVELOPMENTS AND FURTHER RESEARCHES
		2.5.1 Non-Switching Random Impulse Noise Cancellation
		2.5.2 De-Noise Feature Extraction
		2.5.3 Strong Noisy Image De-Noising
		2.5.4 Classify Noise Filtering Results in Seismic Images
	REFERENCES
Chapter 3◾Image De-Blurring
	3.1 OVERVIEW OF IMAGE DE-BLURRING
		3.1.1 General Image Degradation Model
		3.1.2 Blurring Degradation
		3.1.3 Blur Kernel Estimation
			3.1.3.1 Using the Image Observation to Estimate the Blur Function
			3.1.3.2 Using the Point Source Image Experiment to Estimate the Blur Function
			3.1.3.3 Using Blur Modeling to Estimate the Blur Function
	3.2 IMAGE RESTORATION AND DE-BLURRING
		3.2.1 Inverse Filtering
			3.2.1.1 Unconstrained Restoration
			3.2.1.2 Inverse Filtering Principle
			3.2.1.3 Restore Transfer Function
			3.2.1.4 Fast Decomposition Calculation
		3.2.2 Wiener Filtering
			3.2.2.1 Constrained Restoration
			3.2.2.2 Wiener Filter
		3.2.3 Constrained Least Squares Restoration
		3.2.4 Interactive Restoration
	3.3 ESTIMATING MOTION BLUR KERNEL
		3.3.1 Fast Blind De-Convolution
			3.3.1.1 Basic Idea
			3.3.1.2 Blind De-Convolution Process
			3.3.1.3 Calculation Details
		3.3.2 CNN-Based Method
			3.3.2.1 Principle and Process
			3.3.2.2 Network Structure
			3.3.2.3 Network Training
			3.3.2.4 Fuzzy Kernel and Clear Image Estimation
	3.4 LOW-RESOLUTION IMAGE DE-BLURRING
		3.4.1 Network Structure
			3.4.1.1 Loss Function
			3.4.1.2 Generator Network
			3.4.1.3 Discriminator Network
		3.4.2 Loss Function Design and Effects
			3.4.2.1 Various Loss Functions
			3.4.2.2 Combined Final Loss Function
			3.4.2.3 Comparison of Loss Functions
		3.4.3 Multi-Class GAN
			3.4.3.1 Basic Principle
			3.4.3.2 Experimental Effect
	3.5 SOME RECENT DEVELOPMENTS AND FURTHER RESEARCH
		3.5.1 Various De-Blurring Approaches
			3.5.1.1 Hybrid Regularization
			3.5.1.2 Motion De-Blurring Based on Dense Nets
			3.5.1.3 Motion De-Blurring Using Deep Learning-Based Intelligent Systems
		3.5.2 Treating Blurred Images in Applications
			3.5.2.1 Vehicle Logo Recognition
			3.5.2.2 Adversarial Attack Detection
	REFERENCES
Chapter 4◾Image Repairing
	4.1 IMAGE REPAIRING OVERVIEW
		4.1.1 Discrimination and Analysis of Image Repairing
			4.1.1.1 Discrimination
			4.1.1.2 Image Repairing Example
		4.1.2 Principle of Image Repairing
		4.1.3 Image Inpainting for Small-Scale Repairing
			4.1.3.1 Total Variation Model
			4.1.3.2 Mixed Model
		4.1.4 Image Completion for Large-Scale Repairing
			4.1.4.1 Basic Idea
			4.1.4.2 Basic Method Based on Sample
	4.2 ALGORITHMS COMBINED WITH SPARSE REPRESENTATION
		4.2.1 Principle of Sparse Representation
			4.2.1.1 Sparse Representation
			4.2.1.2 Sparse Coefficient
		4.2.2 Basic Sparse Representation Algorithm
			4.2.2.1 Algorithm Steps
			4.2.2.2 Algorithm Key Points
		4.2.3 Improvements to the Sparse Representation Algorithm
	4.3 WEIGHTED SPARSE NON-NEGATIVE MATRIX FACTORIZATION
		4.3.1 Weighted Non-Negative Matrix Factorization
		4.3.2 Filling Algorithm
		4.3.3 WSNMF Based on EM process
	4.4 CONTEXT-DRIVEN HYBRID ALGORITHM
		4.4.1 Overall Flowchart
		4.4.2 Pre- Processing Step
		4.4.3 Sample-Based Repairing Step
		4.4.4 Diffusion-Based Repairing Step
	4.5 SOME RECENT DEVELOPMENTS AND FURTHER RESEARCH
		4.5.1 Categorization of Repairing Methods
		4.5.2 AE and GAN in Image Repairing
			4.5.2.1 AE-Based Techniques
			4.5.2.2 GAN-Based Techniques
			4.5.2.3 Hybrid Techniques
			4.5.2.4 Comparison
	REFERENCES
Chapter 5◾Image De-Fogging
	5.1 SUMMARY OF IMAGE DE-FOGGING APPROACHES
		5.1.1 Methods Based on Image Enhancement
		5.1.2 Methods Based on Image Restoration
	5.2 DCP DE-FOGGING ALGORITHM
		5.2.1 Atmospheric Scattering Model
		5.2.2 DCP Model
		5.2.3 Some Practical Problems
	5.3 IMPROVEMENT IDEAS AND TECHNIQUES
		5.3.1 Determination of Global Atmospheric Light Region
			5.3.1.1 Determine the Atmospheric Light Estimation Point Based on the Physical Meaning
			5.3.1.2 Calculate the Densest Fog Region with the Help of a Quadtree
		5.3.2 Global Atmospheric Light Value Correction
			5.3.2.1 Global Atmospheric Light Value Weighted Correction
			5.3.2.2 Atmospheric Light Color Value Correction
		5.3.3 Scale Adaptation
			5.3.3.1 Initial Scale Based on the Color Characteristics
			5.3.3.2 Correct the Scale According to the Edge Features
		5.3.4 Atmospheric Transmittance Estimation
			5.3.4.1 Fuse Dark Channel Values to Estimate Atmospheric Transmittance
			5.3.4.2 Refining Atmospheric Transmittance Based on Local Adaptive Wiener Filtering
		5.3.5 Dense Foggy Image De-Fogging
			5.3.5.1 Algorithm Flowchart
			5.3.5.2 Fog Density Factor Estimation
			5.3.5.3 Atmospheric Light Estimation Based on Guided Filtering
	5.4 INTEGRATED ALGORITHM FOR REDUCING DISTORTION
		5.4.1 Algorithm Flowchart
		5.4.2 T Space Conversion
		5.4.3 Atmospheric Scattering Map in Transmittance Space
		5.4.4 Sky Region Detection
		5.4.5 Contrast Enhancement
	5.5 EVALUATION OF DE-FOGGING EFFECTS
		5.5.1 Objective Evaluation Index
			5.5.1.1 No Reference Evaluation Index
			5.5.1.2 Visible Edge Gradient
			5.5.1.3 Visual Perception Computing
		5.5.2 Examples of Evaluations Combining Subjective and Objective Indices
			5.5.2.1 Evaluation Indicators and Calculations
			5.5.2.2 Experiments and Results
	5.6 SOME RECENT DEVELOPMENTS AND FURTHER RESEARCH
		5.6.1 Nighttime Fog Removal
			5.6.1.1 A Photographic Negative Imaging Inspired Method
			5.6.1.2 Combining Bright and DCP
			5.6.1.3 Deep Learning for Image De-Fogging
		5.6.2 More General Fog Removal Techniques
			5.6.2.1 Non-Learning Image De-Fogging Techniques
			5.6.2.2 Learning-Based Image De-Fogging Techniques
	REFERENCES
Chapter 6◾Image Reconstruction from Projection
	6.1 PROJECTION RECONSTRUCTION FORMS
		6.1.1 Transmission Tomography
			6.1.1.1 CT Value
			6.1.1.2 CT System
		6.1.2 Emission Tomography
			6.1.2.1 Positron Emission CT
			6.1.2.2 Single-Photon Emission CT
		6.1.3 Reflection Tomography
		6.1.4 Electrical Impedance Tomography
		6.1.5 Magnetic Resonance Imaging
	6.2 PRINCIPLES OF RECONSTRUCTION FROM PROJECTION
		6.2.1 Basic Model
		6.2.2 Radon Transform
	6.3 INVERSE FOURIER TRANSFORM RECONSTRUCTION
		6.3.1 The Basic Steps and Definitions
		6.3.2 Fourier Transform Projection Theorem
		6.3.3 Model Reconstruction
	6.4 BACK-PROJECTION RECONSTRUCTION
		6.4.1 Principles of Back-Projection Reconstruction
		6.4.2 Convolutional Back-Projection Reconstruction
			6.4.2.1 Continuous Formula Derivation
			6.4.2.2 Discrete Calculation
			6.4.2.3 Reconstruction from Fan-Beam Projection
			6.4.2.4 Comparison of Inverse Fourier Transform Reconstruction Method and Convolutional Back-Projection Reconstruction Method
		6.4.3 Other Back-Projection Reconstruction Methods
			6.4.3.1 Back-Projection Filtering
			6.4.3.2 Filtered Back-Projection
	6.5 ITERATIVE RECONSTRUCTION
		6.5.1 Iterative Reconstruction Model
		6.5.2 Algebraic Reconstruction Technique
			6.5.2.1 Basic Algorithm
			6.5.2.2 Relaxed Algebraic Reconstruction Technique
			6.5.2.3 Simultaneous Algebraic Reconstruction Technology
			6.5.2.4 Some Characteristics of the Series Expansion Technique
		6.5.3 Maximum Likelihood-Maximum Expectation Reconstruction Algorithm
	6.6 COMBINED RECONSTRUCTION
	6.7 SOME RECENT DEVELOPMENTS AND FURTHER RESEARCH
		6.7.1 Metal Artifact Reduction
			6.7.1.1 Metal Artifact
			6.7.1.2 Classical MAR Methods
			6.7.1.3 Deep Learning-Based MAR Methods
		6.7.2 4-D Cone-Beam CT Reconstruction
			6.7.2.1 Cone-Beam CT Reconstruct ion
			6.7.2.2 4-D Cone-Beam CT
			6.7.2.3 4-D Cone-Beam CT Reconstruction
	REFERENCES
Chapter 7◾Image Watermarking
	7.1 OVERVIEW OF WATERMARKING
		7.1.1 Embedding and Detection of Watermark
		7.1.2 Watermark Characteristics
			7.1.2.1 Saliency
			7.1.2.2 Robustness
			7.1.2.3 Security
			7.1.2.4 Other Characteristics
		7.1.3 Watermark Classification
			7.1.3.1 Publicity Classification
			7.1.3.2 Perceptual Classification
			7.1.3.3 Meaning/Content Classification
	7.2 WATERMARK MEASUREMENT INDEX
		7.2.1 Saliency/Perception Measurement
			7.2.1.1 Perception Benchmark Metrics
			7.2.1.2 Objective Distortion Metrics
		7.2.2 Robustness Measurement
		7.2.3 Security and Watermark Attack
			7.2.3.1 Attack type
			7.2.3.2 Typical Attack Examples
			7.2.3.3 Watermark Attack Analysis
	7.3 DCT DOMAIN WATERMARK
		7.3.1 Features and Principles
		7.3.2 Meaningless Watermarking Algorithm
			7.3.2.1 Watermark Embedding
			7.3.2.2 Watermark Detection
			7.3.2.3 Watermark Performance
		7.3.3 Meaningful Watermarking Algorithm
			7.3.3.1 Watermark Design
			7.3.3.2 Watermark Embedding
			7.3.3.3 Watermark Detection
			7.3.3.4 The Robustness of Watermark
	7.4 DWT DOMAIN WATERMARK
		7.4.1 Features and Process
		7.4.2 Human Visual Characteristics
		7.4.3 Wavelet Watermarking Algorithm
			7.4.3.1 Watermark Embedding
			7.4.3.2 Watermark Detection
			7.4.3.3 Wavelet Domain Watermark Performance Test
	7.5 SOME RECENT DEVELOPMENTS AND FURTHER RESEARCH
		7.5.1 Zero-Watermarking
			7.5.1.1 Basic Principle
			7.5.1.2 Watermark Generation
			7.5.1.3 Watermark Verification
			7.5.1.4 Feature Extraction: Higher-Order Statistics
			7.5.1.5 Feature Extraction: Singular Values and Extension
			7.5.1.6 Video Zero-Watermarking
			7.5.1.7 Video Zero-Watermark Based on CNN and a Self-Organizing Map
		7.5.2 More Extensive Watermarking Technology
			7.5.2.1 Database Watermarking
			7.5.2.2 3-D Mesh Watermarking
			7.5.2.3 Bio-Medical Signal Data Watermarking
			7.5.2.4 Watermarking in Different Application Domains
	REFERENCES
Chapter 8◾Image Super-Resolution
	8.1 PRINCIPLE OF IMAGE SR
		8.1.1 Basic Model and Technology Classification
			8.1.1.1 Image Observation Model
			8.1.1.2 SR Technology Identification
			8.1.1.3 Technology Classification
		8.1.2 SR Restoration Based on Single Image
			8.1.2.1 Image Enlargement
			8.1.2.2 SR Restoration
		8.1.3 SR Reconstruction Based on Multiple Images
			8.1.3.1 Typical Method
			8.1.3.2 Video SR
	8.2 SR TECHNOLOGY BASED ON LEARNING
		8.2.1 Conventional Process
		8.2.2 Example-Based Single-Frame SR
			8.2.2.1 Basic Principles and Steps
			8.2.2.2 Training Set Generation
			8.2.2.3 Markov Network Algorithm
			8.2.2.4 Single-Pass Algorithm
			8.2.2.5 The Matching of Image Patches
		8.2.3 Example-Based Multi-Frame SR
			8.2.3.1 The Overall Process
			8.2.3.2 Specific Key Points
		8.2.4 Methods Combined with Total Variation Regularization
		8.2.5 Learning-Based Methods
	8.3 SR RECONSTRUCTION BASED ON SPARSE REPRESENTATION
		8.3.1 Reconstruction Process
		8.3.2 Sparse Coding
		8.3.3 Dictionary Learning
		8.3.4 Image Reconstruction
	8.4 SR RECONSTRUCTION BASED ON LOCALLY CONSTRAINED LINEAR CODING
		8.4.1 Locally Constrained Linear Coding
		8.4.2 SR Reconstruction Algorithm Based on Locally Constrained  Linear Coding
		8.4.3 Multi-Frame Image SR Reconstruction
		8.4.4 Reconstruction Results and Method Comparison
	8.5 SOME RECENT DEVELOPMENTS AND FURTHER RESEARCH
		8.5.1 Overview of SR Based on Deep Learning
		8.5.2 Loss Functions and Evaluation Indicators
			8.5.2.1 Loss Functions
			8.5.2.2 Evaluation Indicators
	REFERENCES
INDEX