摘要
针对极化合成孔径雷达分类过程中较难同时获得精确的边缘和光滑的同质区域的问题,提出了一种基于Wishart距离的高概率选择分类器与自适应马尔科夫随机场相结合的分类方法,对极化合成孔径雷达图像分类.首先,将Wishart分类器应用于概率输出的支撑矢量机中,根据高概率选择得到一个基于像素的初始分类结果,并将此结果结合不同的边缘检测方法得到一个精确的边缘;其次,采用自适应窗口的马尔科夫随机场对上一步的分类结果进行修正,该过程在得到平滑区域的同时,也保持了上一步分类结果的边缘.实验结果表明,该算法提高了极化合成孔径雷达图像分类的精度,并保持了图像的细节信息.
It is difficult to obtain the accurate boundaries and the smooth regions for polarimetric SAR image classification.In order to solve this problem,a novel classification scheme is proposed that combines the Wishart-based high-probabilistic support vector machine(SVM)and adaptive markov random fields(MRF).First,a Wishart classifier is applied with the probabilistic SVM,according to high-probalistic selection,an initial pixel-based classification result is obtained.Then by combining this result with other edge detection methods,it can access the accurate boundaries.Second,adaptive MRF is proposed based on the edge of the image to further revise the previous classification.In this way,smooth regions are obtained and accurate boundaries are maintained simultaneously.Experimental results show that the proposed method improves the classification performance and that details of the image are preserved.
出处
《西安电子科技大学学报》
EI
CAS
CSCD
北大核心
2017年第6期59-64,共6页
Journal of Xidian University
基金
国家自然科学基金资助项目(61671350)
