Abstract: The speckle noise of high resolution airborne MiniSAR image is serious, and in image the edge of the reclamation is fuzzy and messy. For more efficient extraction of reclamation information, a method of watershed segmentation was presented based on texture feature for SAR image of reclamation. First, the airborne MiniSAR image was filtered by the gray level co-occurrence matrix, and the texture feature image was obtained. Then the watershed algorithm was used to segment the texture feature image and to get the morphological reconstruction image, and finally, the threshold method was used to segment the morphological image, and to get the final results of the two values. On the one hand, this method suppresses the influence of speckle noise by adjusting the gray co-occurrence matrix texture filtering window size . On the other hand, it uses watershed algorithm of fuzzy edge clutter images to improve the accuracy of information extraction of sea reclamation. Test results show that this method is effective for the image segmentation of high resolution SAR image.

Key words: airborne MiniSAR, reclamation, gray level co-occurrence matrix, watershed, image segmentation