Adaptive Image Processing: A Computational Intelligence Perspective

Adaptive image processing is one of the most important techniques in visual information processing, especially in early vision such as image restoration, filtering, enhancement, and segmentation. While existing books present some important aspects of the issue, there is not a single book that treats this problem from a viewpoint that is directly linked to human perception - until now. This reference treats adaptive image processing from a computational intelligence viewpoint, systematically and successfully, from theory to applications, using the synergies of neural networks, fuzzy logic, and evolutionary computation. Based on the fundamentals of human perception, this book gives a detailed account of computational intelligence methods and algorithms for adaptive image processing in regularization, edge detection, and early vision. Adaptive Image Processing: A Computational Intelligence Perspective consists of 8 chapters: Chapter 1 - Provides material of an introductory nature to describe the basic concepts and current state-of-the-art in the field of computational intelligence for image restoration and edge detection Chapter 2 - Gives a mathematical description of the restoration problem from the neural network perspective, and describes current algorithms based on this method Chapter 3 - Extends the algorithm presented in chapter 2 to implement adaptive constraint restoration methods for both spatially invariant and spatially variant degradations Chapter 4 - Utilizes a perceptually motivated image error measure to introduce novel restoration algorithms Chapter 5 - Examines how model-based neural networks can be used to solve image restoration problems Chapter 6 - Probes image restoration algorithms, making use of the principles of evolutionary computation Chapter 7 - Explores the difficult concept of image restoration when insufficient knowledge of the degrading function is available Chapter 8 - Studies the subject of edge detection and characterization using model-based neural networks The first to treat adaptive image processing from a computational intelligence perspective, this work provides an excellent reference in R&D practice to researchers and IT technologists, is most suitable for teaching image processing and applied neural network courses, and will be of equal value for technical managers and executives in industries where intelligent visual information processing is required.

Table of Contents:
PREFACE INTRODUCTION The Importance of Vision Adaptive Image Processing The Three Main Image Feature Classes Difficulties in Adaptive Image Processing System Design Computational Intelligence Techniques Scope of the Book Contributions of the Current Work Overview of this Book FUNDAMENTALS OF NEURAL NETWORK IMAGE RESTORATION Image Distortions Image Restoration Neural Network Restoration Algorithms in the Literature An Improved Algorithm Analysis Implementation Considerations A Numerical Study of the Algorithms Summary SPATIALLY ADAPTIVE IMAGE RESTORATION Introduction Dealing with Spatially Variant Distortion Adaptive Constraint Extension of the Penalty Function Model Correcting Spatially Variant Distortion Using Adaptive Constraints Semi-Blind Restoration Using Adaptive Constraints Implementation Considerations More Numerical Examples Adaptive Constraint Extension of the Lagrange Model Summary PERCEPTUALLY MOTIVATED IMAGE RESTORATION Introduction Motivation A LVMSE-Based Cost Function A Log LVMSE-Based Cost Function Implementation Considerations Numerical Examples Summary MODEL-BASED ADAPTIVE IMAGE RESTORATION Model-Based Neural Network Hierarchical Neural Network Architecture Model-Based Neural Network with Hierarchical Architecture (HMBNN) HMBNN for Adaptive Image Processing The Hopfield Neural Network Model for Image Restoration Adaptive Regularization - An Alternative Formulation Regional Training Set Definition Determination of the Image Partition The Edge-Texture Characterization (ETC) Measure The ETC Fuzzy HMBNN for Adaptive Regularization Theory of Fuzzy Sets Edge-Texture Fuzzy Model Based on ETC Measure Architecture of the Fuzzy HMBNN Estimation of the Desired Network Output Fuzzy Prediction of Desired Gray Level Value Experimental Results Conclusion ADAPTIVE IMAGE REGULARIZATION USING EVOLUTIONARY COMPUTATION Introduction Introduction to Evolutionary Computation The ETC-pdf Image Model Adaptive Regularization Using Evolutionary Programming Experimental Results Other Evolutionary Approaches for Image Restoration Summary BLIND IMAGE DECONVOLUTION Introduction Computational Reinforced Learning Soft-Decision Method Simulation Examples Conclusions EDGE CHARACTERIZATION USING MODEL-BASED NEURAL NETWORKS Introduction MBNN Model for Edge Characterization Network Architecture Training Stage Recognition State Experimental Results Summary