Classification of SAR image based on gray cooccurrence matrix and support vector machine

In this paper, the classification method of SAR image based on support vector machine (SVM) with extracting the gray feature, texture feature (gray cooccurrence matrix) was proposed. Compared the results of different kernel functions with the result of maximum likelihood classifier, this approach was proved to be able to classify those patterns that can´t be distinguished exactly by the maximum likelihood classifier. On the other hand, experimental results showed the classification precision of SVM with linear kernel function is higher than that with Gauss kernel function.