A New Automatic Ship Detection Method Using $L$-Band Polarimetric SAR Imagery

Ship-sea contrast can be improved significantly when the full polarimetric information is used, compared with the information provided by a single polarization channel. Therefore, a new automatic ship detection method, termed SPAN Wishart (SPWH), is proposed in this paper based on an unsupervised classification concept, which combines the SPAN of a polarimetric SAR (POLSAR) data with the complex Wishart classifier. The significant improvement of this technique is to utilize the SPAN of ship cluster center as an iterative termination criterion to realize the automatic ship detection. Then, another method based on multifrequency is proposed to discriminate between ships and their ambiguities to provide a substitute for the ground truth for the subsequent validation of the SPWH algorithm by using an AIRSAR polarimetric dataset, which consists of C- and L-band data covering Kojimawan in Tamano of Japan. After that, the SPWH is validated only using the L-band data with the defined ships by comparing the performance of the SPWH with a traditional CFAR detector. By contrast, the SPWH algorithm is more effective, robust, and completely automatic.