基于融合距离的极化SAR图像非局部均值滤波  

Nonlocal means filter for polarimetric SAR images based on fusion distance

作　　者：曾顶 殷君君[1] 杨健[2] ZENG Ding;YIN Junjun;YANG Jian(School of Computer and Communication Engineering,University of Science and Technology Beijing,Beijing 100083,China;Department of Electronic Engineering,Tsinghua University,Beijing 100084,China)

机构地区：[1]北京科技大学计算机与通信工程学院,北京100083 [2]清华大学电子工程系,北京100084

出　　处：《系统工程与电子技术》2024年第5期1493-1502,共10页Systems Engineering and Electronics

基　　金：国家自然科学基金(62222102,62171023,U20B2062)资助课题。

摘　　要：在极化合成孔径雷达(synthetic aperture radar,SAR)图像降噪领域,常见的非局部均值滤波仅依靠像素间的统计距离进行相似性度量,忽略了像素点的空间信息。本文结合极化SAR数据统计特性和图像空间特征作为像素间的相似性度量,提出了一种利用融合距离来计算相邻窗口权重的方法——基于融合距离的非局部均值滤波器。融合距离的引入使得滤波器能够更全面的评估像素间的相似性,从而得到更合适的像素权重。此外,本方法还引进变异系数对邻域窗口的权重进行评估,通过该参数可以控制滤波的程度。在多幅极化SAR图像上的实验结果表明,所提出的滤波器能够在有效抑制斑点噪声的同时保留较为完整的图像边缘信息和极化散射特性。In the field of polarimetric synthetic aperture radar(SAR)image denoising,the common nonlocal means(NLM)filter only relies on the statistical distance between pixels to measure the similarity and ignores the spatial information of them.This study combines the statistical characteristics of polarimetric SAR data and image spatial features as similarity measures between pixels,and proposes a method for calculating adjacent window weights using fusion distance,which names NLM filter based on fusion distance(FD-NLM).The introduction of fusion distance enables the filter to comprehensively evaluate the similarity between pixels,thereby obtaining more appropriate pixel weights.In addition,this method also employs the coefficient of variation(CV)to evaluate the weight of neighborhood windows,and using this parameter to control the filtering degree.The experimental results on multiple polarimetric SAR images show that the proposed filter can effectively suppress speckle noise while retaining relatively complete image edge information and polarization scattering characteristics.

关 键 词：极化合成孔径雷达 非局部均值滤波 相似性度量 变异系数 

分 类 号：TN958[电子电信—信号与信息处理]