A Precise and Fast SAR Image Segmentation Method  

作　　者：JU Yanwei TIAN Zheng ZHANG Yan 

机构地区：[1]Department of Applied Mathematics, Northwestern Polytechnical University, Xi＇an 710072, China [2]The 14th Research Institute, CETC, Nanjing 210013, China

出　　处：《Chinese Journal of Electronics》2007年第3期471-475,共5页电子学报（英文版）

基　　金：This work is supported by the National Natural Science Foundation of China （No.60375003） and Aeronautics and Astronautics Basal Science Foundation of China （No.03153059）.

摘　　要：A Generalized multiresolution likelihood ratio （GMLR）, which can increases the distinction between different signals by fusing their more features, is defined. The GMLR for SAR （Synthetic aperture radar） image, the features of it which produced by the pyramid representation of SAR imagery characterizes and exploits the multiscale stochastic structure inherent in SAR imagery due to radar speckle. In our unsupervised SAR image segmentation method, a Spatially variant mixture multiscale autoregressive prediction （SVMMARP） model is proposed to estimate the parameters of GMLR based on maximum likelihood estimation. In order to satisfy the independence assumption of maximum likelihood estimation and reduce the segmentation time greatly, we perform our method based on the Bootstrap sampling technique. The algorithm avoids some drawbacks that existed in some popular segmentation techniques. Experimental results demonstrate that our algorithm performs fairly well.

关 键 词：Generalized multiresolution likelihood ratio （GMLR） Bootstrap sampling Spatially variant mixture multiscale autoregressive prediction （SVMMARP） model Maximization likelihood estimation SAR image Unsupervised segmentation 

分 类 号：TN03[电子电信—物理电子学]