The image segmentation is one of the most used steps in image processing applications. Indeed, this process enables us to extract feature vectors which could be useful for experts decision and recognition task. This paper deals with the radar images segmentation for recognition targets. However, the radar images, generally include noise resulting from the acquisition step and from the several techniques of radar image reconstruction. In this work, we aimed to extract the general shape from the ISAR images. Accordingly, we combined two techniques: the Fisher algorithm and the watershed technique.