Chapter 5 surveys some representative computational approaches to texture analysis and synthesis. In particular, basic concepts, theories and techniques related to MGRF-based texture modelling are introduced, followed by reviews of two representative MGRF models in detail. Also, today's main texture synthesis methods are reviewed, with focus on a family of fast texture synthesis techniques based on non-parametric sampling.
Chapter 6 investigates the problem of identification of a generic MGRF model of homogeneous textures in detail. Both the conventional identification and the new structural approach are covered.
Chapter 7 introduces a fast synthesis technique called bunch sampling derived from a texel-based texture description.The synthesis algorithm directly transfers texels randomly sampled from a training texture into the synthetic texture in line with derived placement rules. Experimental results of the synthesis for a broad range of homogeneous textures are also presented.
Finally, Chapter 8 summarises and concludes the thesis. Possible directions of future researches are also discussed.