Abstract:

It is difficult to extract smooth contours from inverse synthetic aperture radar (ISAR) images because their low resolution makes them sparse in image domain. The ISAR images with noise and clutter are modeled first. Then a method separating objects’contours from the background in ISAR images is presented. The method consists of a morphology-based contour extraction algorithm of ISAR images and a contour smoothing process based on wavelet analysis. The experiment’s results show that the method avoids the difficulty introduced by the ISAR images’sparsity, which makes it hard to get connected and smooth contours. It gives better results than some other contour extraction methods and is suitable for extracting objects’smooth contours from ISAR images. ISAR images can be better used for feature extraction and automatic recognition after extracting the smooth contours through image domain analysis.