Next:
Abstract
Texture Analysis and Synthesis using a Generic Markov-Gibbs Image Model
Dongxiao Zhou
Abstract
Acknowledgments
Declaration
Introduction
What is a Texture?
Human Texture Perception
Texture Analysis and its Applications
Texture Classification
Texture Segmentation
Shape from Texture
Texture Synthesis
Motivation
Organisation of the Thesis
Notational Conventions
Texture Analysis and Synthesis: Related Works
Texture Analysis: an Overview
Descriptive Approach
Generic Approach
Markov-Gibbs Random Field Models of Textures
Images and Probabilities
Markov-Gibbs Random Fields
Markov Random Field: the Definition
Gibbs Distribution
Markov-Gibbs Equivalence
Texture Modelling Based on MGRFs
Model Identification
Maximum Entropy Principle
Maximum Likelihood Estimation
Parameter Estimation via MCMC
Auto-models
FRAME model
Today's Methods of Texture Synthesis
Pyramid-Based Texture Synthesis
Multi-resolution Sampling
Pixel-based Non-parametric Sampling
Block Sampling
Cut-Primed Smart Copying
Summary
Generic Markov-Gibbs Model and its Identification
Introduction
Model Parameters and Sufficient Statistics
Model Identification
Analytic First Approximation of Potentials
Characteristic Neighbourhood
Potential Refinement
Structural Identification of a Generic MGRF Model
Model-Based Interaction Map
Characteristic Neighbourhood
Geometric Structure of Texels
Placement Rule for Texels
Comparison and Discussion
Summary
Texture Synthesis by Bunch Sampling
Texture Synthesis by a Generic MGRF Model
Bunch Sampling
Algorithm Details
Implementation Issues
Experimental Results
Synthesis of Homogeneous Textures
Synthesis of Weakly-Homogeneous Textures
Comparison with Related Synthesis Methods
Summary
Conclusion and Future Work
Contribution
Limitations
Future Work
Conclusion
Approximate Potential Estimates
Synthesised Natural Textures
Bibliography
About this document ...
dzho002 2006-02-22
