A New Fast SAR Image Segmentation Algorithm Based on Grid Coding and Region Merging

A new SAR image partition model is constructed based on 8-neighbor grid code, which is fast solved by region merging. Utilizing multi-direction ratio edge detector to construct Ratio Edge Strength Map (RESM) of SAR image, a novel thresholding method is proposed to suppress the minima value in the homogeneous region of RESM, which reduces the number of regions in an initial partition produced by watershed of the thresholding processed RESM. Sub-optimization of the partition model is obtained by merging adjacent region pair iteratively. Region Adjacency Graph (RAG) and its Nearest Neighbor Graph (NNG) characteristic are used to speed up the proceeding of region merging. Precision (P ) and Recall (R) are introduced to evaluate the boundary localization precision of segmentation methods. Compared with three widely used methods, the proposed method has higher boundary localization precision and lower computational complexity.