Optimal approach to SAR image segmentation and classification

Many algorithms have been proposed for the segmentation and classification of SAR imagery. Typically these are heuristic in basis, and more successful on some types of imagery than others. Methods based on global optimisation techniques can achieve the optimal solution of the posed problem; the algorithm is then characterised by an objective function and the chosen optimisation technique. The authors construct an appropriate objective function for SAR segmentation or classification, and show how to optimally quantify the relationship between the competing terms within it. This removes the previous need for the introduction of heuristics. Also described is an implementation based on simulated annealing, in which considerable attention has been paid to details, and examples are given of the resulting algorithm's performance upon real data.