Aiming at detecting the change regions of high resolution Synthetic Aperture Radar (SAR) images, we propose to use the Dempster-Shafer (D-S) evidence theory to fuse coherent/incoherent features from sensors that form an integral part of the system. First, we use the Simple Linear Iterative Clustering (SLIC) segmentation algorithm to implement multi-scale joint segmentation for multi-temporal SAR images. Second, we extract multiple intensity and coherence difference features on each segment level by SLIC using mean operator to complete the fusion of multi-scale features to get the multi-feature difference mapped by a ratio operator. Finally, we fuse the multi-feature difference maps to get the final change detection result using the D-S evidence theory. The experimental results in our study prove the effectiveness of our proposed computational algorithm.

The National Natural Sciences Foundation of China (61271401, 61461040), and the Program for Young Talents of Science and Technology in Universities of Inner Mongolia Autonomous Region (NJYT-14-B09)