APPLICATIONS OF SIMULATED ANNEALING TO SAR IMAGE CLUSTERING AND CLASSIFICATION PROBLEMS

This paper deals with simulated annealing applications to some unsuper vised classification problems. First, we present an adaptation of the simulated annealing technique to the clustering problem, and compare i ts results with those provided by classical c-mean clustering algorith ms, which lead only to local optimality. It is shown that simulated an nealing technique yields improved results over the standard c-means me thod. Then, we compare the classified images obtained from two differe nt clustering algorithms: simulated annealing, and K-means guided by i nitialization from optical data, applied to MAC-Europe'91 images. The study reveals a poor robustness of SAR image classification versus the estimated cluster characteristics, and shows that most of culture typ es were best identified by the global optimization.