高概率选择和自适应MRF的极化SAR分类
发布时间：2019-05-28 06:09
【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.
[Abstract]:In order to solve the problem that it is difficult to obtain accurate edges and smooth homogeneous regions at the same time in the process of polarimetric synthetic aperture radar (SAR) classification, a classification method based on Wishart distance is proposed, which combines high probability selection classifiers with adaptive Markov random fields. Classification of polarimetric synthetic aperture radar (SAR) images. Firstly, the Wishart classifier is applied to the support vector machine with probabilistic output, and an initial classification result based on pixels is obtained according to the high probability selection, and an accurate edge is obtained by combining the results with different edge detection methods. Secondly, the Markov random field of the adaptive window is used to modify the classification results of the previous step. The process not only obtains the smooth region, but also maintains the edge of the previous classification results. The experimental results show that the algorithm improves the accuracy of polarimetric synthetic aperture radar (SAR) image classification and maintains the details of the image.
【作者单位】： 西安电子科技大学智能感知与图像理解教育部重点实验室;
【基金】：国家自然科学基金资助项目(61671350)
【分类号】：TN957.52
，
本文编号：2486803
[Abstract]:In order to solve the problem that it is difficult to obtain accurate edges and smooth homogeneous regions at the same time in the process of polarimetric synthetic aperture radar (SAR) classification, a classification method based on Wishart distance is proposed, which combines high probability selection classifiers with adaptive Markov random fields. Classification of polarimetric synthetic aperture radar (SAR) images. Firstly, the Wishart classifier is applied to the support vector machine with probabilistic output, and an initial classification result based on pixels is obtained according to the high probability selection, and an accurate edge is obtained by combining the results with different edge detection methods. Secondly, the Markov random field of the adaptive window is used to modify the classification results of the previous step. The process not only obtains the smooth region, but also maintains the edge of the previous classification results. The experimental results show that the algorithm improves the accuracy of polarimetric synthetic aperture radar (SAR) image classification and maintains the details of the image.
【作者单位】： 西安电子科技大学智能感知与图像理解教育部重点实验室;
【基金】：国家自然科学基金资助项目(61671350)
【分类号】：TN957.52
，
本文编号：2486803
本文链接：https://www.wllwen.com/kejilunwen/xinxigongchenglunwen/2486803.html
