Detection of River in SAR Images

A method of river detection for SAR images is put forward by this paper. It uses kernelized fuzzy c-means (KFCM) algorithm to realize initial segmentation of river in original SAR images. It utilizes mathematical morphology filtering. It removes false alarm targets and retains geometric features of images. In order to further remove the false alarm targets, we use a criterion of the water body. Experimental result indicates that the algorithm can detect river effectively, and so, has significant practical value.