全极化SAR图像边缘检测的随机距离法  

Edge Detection of PolSAR Image Based on Stochastic Distance

作　　者：王庆[1] 曾琪明[1] 张海真[1] 焦健[1] 

机构地区：[1]北京大学遥感与地理信息系统研究所北京市空间信息集成与3S工程应用重点实验室,北京100871

出　　处：《测绘学报》2015年第7期753-760,共8页Acta Geodaetica et Cartographica Sinica

基　　金：国家863计划(2012AA121304);国家自然科学基金(41171267)~~

摘　　要：提出了一种基于复Wishart分布随机距离的PolSAR图像边缘检测方法,将统计学中的随机距离理论引入PolSAR图像边缘检测中,依据的主要原理是边缘两侧类别之间随机距离的大小与边缘的方向和两侧类别差异的高度相关性。通过模拟数据对随机距离检测边缘的性能进行了全面分析,表明随机距离具有准确的边缘定向和定位能力。并利用基于复Wishart分布随机数生成器模拟的PolSAR图像和一景机载全极化SAR图像进行了验证。试验结果证实了该方法检测边缘的效果。A new edge detection methodology in PolSAR images is proposed, which is based on stochastic distance in the statistical theory and combined with complex Wishart distribution. Its main principle is inspired from the phenomenon that stochastic distance of two classes separated by an edge is closely related to the edge direction and the contrast of two classes. Simulation experiments are carried out to analyze the performance of the proposed methods. Results prove that methods have better capabilities in edge orientation and edge positioning than common used methods. Then, the proposed detection procedure is tested on a simulated PolSAR image with the complex Wishart distribution, as well as an airborne fully polarimetric SAR image.

关 键 词：随机距离 WISHART分布 POLSAR 边缘检测 

分 类 号：P237[天文地球—摄影测量与遥感]