Extraction of inshore vessels in very high resolution SAR images
Guangjiao Zhou, Ye Zhang, Qi Li · 2017
In very high resolution (VHR) SAR images, traditional method failed to accurately and completely extract vessels because of the holes and fractures in the target body in addition to the interference of adjacent objectives. In this paper, a saliency method based on SLIC (simple linear iteration clustering) segmentation is adopted to detect the targets, and the salient results are further post-processing to remove the false alarms and fill the blanks in the target body. Experimental results on Terra-SAR images demonstrate that the proposed approach can maintain the integrity of vessel and decrease the false alarm rate.