The goal of this research project is the development of stochastic models and statistically optimal segmentation algorithms for noisy and textured images. Specific objectives include: further development of hierarchical Gibbs distribution or Markov random field models to represent real images; development of models for speckle corrupted synthetic aperture radar images; investigation of the relationships between Markov-type models and connections between Markov random field models and robust processing; development of estimators for the parameters in hierarchical image models; and development of statistically optimal segmentation algorithms for images modelled with hierarchical Gibbs distribution models. For parameter estimation and segmentation, relaxation methods will be used. The objective of this research is to develop algorithms for segmenting real world images. Results of this research could be important in fields of synthetic aperture radar and image interpretation.
