Abstract—Discrimination of farmland areas on different row directions from SAR image is very significant to the selection of the optimal row direction for the cultivation and the yield. In order to reduce the effects of the speckles and improve the accuracy, the paper put forward a method which extracts the farmland areas on different row directions with region growing algorithm based on gray-level co-occurrence matrix. Particularly, the method compounds three best GLCM texture feature images and transforms them using HIS algorithm and then uses the intensity image of HIS results for region growing algorithm in order to extract the farmland areas on different row directions. The extraction results with the method in this paper are proved to be quite satisfactory.

Key words—SAR image, row direction, texture feature, region growing algorithm.