基于Contourlet域HMT和D-S证据融合的SAR图像分割  

Segmentation Algorithm for SAR Images Based on Fusion of HMT in the Contourlet Domain and D-S Theory of Evidence

作　　者：吴艳[1] 焦惊眉 杨晓丽[1] 肖平 李明[3] 

机构地区：[1]西安电子科技大学电子工程学院,陕西西安710071 [2]陕西测绘局,陕西西安710054 [3]西安电子科技大学雷达信号处理国家重点实验室,陕西西安710071

出　　处：《测绘学报》2011年第2期148-155,共8页Acta Geodaetica et Cartographica Sinica

基　　金：国家自然科学基金(60872137);国防预研基金(9140C010307100C01);航空科学基金(20080181002)

摘　　要：针对Contourlet的多尺度、局部化、方向性和各向异性等优点,结合隐马尔科夫树(hidden Markov tree,HMT)模型和D-S(Dempster-Shafer)证据理论,提出一种新的SAR图像分割算法。该算法首先将隐马尔科夫树模型推广到Contourlet域,在多尺度HMT上采用D-S证据融合理论有效地融合Contourlet系数的持续性和聚集性,最后导出融合后的最大后验多尺度分割公式。本文算法对实测SAR图像进行分割试验,试验结果表明:与小波域上的HMT-MRF(Markovrandom field,MRF)融合分割及Contourlet域上HMT和MRF分割算法相比,本文算法在抑制斑点噪声的同时,可有效地提高SAR图像的分割精度。Considering the Contourlet＇s advantages of multiscale,localization,directionality and anisotropy,a new segmentation algorithm is proposed for SAR images based on hidden Markov tree in the Contourlet domain and D-S theory of evidence.Firstly,the algorithm extends the hidden Markov tree framework to the Contourlet domain.Then,the clustering and persistence of the Contourlet transform are effectively fused using D-S theory in HMT model,and the maximum a posterior（MAP） segmentation equation for the new fusion model is deduced in the end.The algorithm has been used to segment the real SAR images,and experimental results show that the proposed algorithm effectively reduces the influence of multiplicative noises,improves the segmentation accuracy and provides a better visual quality for SAR images over the algorithms based on HMT-MRF in the wavelet domain,HMT and MRF in the Contourlet domain,respectively.

关 键 词：SAR图像分割 CONTOURLET变换 隐马尔科夫树(HMT) D-S证据理论 

分 类 号：P237[天文地球—摄影测量与遥感]