A Ship Target Detection Method For Sar Image Based On Local Region

Using the knowledge of image processing, machine learning and pattern recognition to develop the techniques of acquiring target regions from Synthetic Aperture Radar (SAR) images accurately has caught researchers' wide attention. A ship target detection method for SAR image based on local region under the multi-scale and multi-target condition is introduced in this paper. This method firstly learns the gradient-based local region generation mode based on the training data with object annotation, then generates a small number of local regions with different sizes, and finally uses the local region based CFAR detector, where the extracted object regions are regarded as the guard windows instead of setting fixed guard window to detect the true object regions. Due to the introduction of local regions, the proposed method can obtain good detection performance in the multi-scale situation, and directly obtain the accurate target regions to avoid the problem caused by the target clustering in traditional ship detection method. The effectiveness of the proposed method is verified using measured RADASAT-2 data by comparing with the traditional SAR ship target detection methods.