一种改进三维块匹配的SAR图像去噪方法  

An improved BM3D algorithm for SAR image denoising

作　　者：刘雪晴 卢小平[1] 马靓婷 LIU Xueqing;LU Xiaoping;MA Liangting(Key Laboratory of Spatio-temporal Information and Ecological Restoration of Mines,MNR,Henan Polytechnic University,Jiaozuo,Henan 454003,China)

机构地区：[1]河南理工大学自然资源部矿山时空信息与生态修复重点实验室,河南焦作454003

出　　处：《测绘科学》2021年第7期115-119,144,共6页Science of Surveying and Mapping

基　　金：2016国家重点研发计划项目(2016YFC0803103)。

摘　　要：针对BM3D算法对合成孔径雷达图像的乘性相干斑噪声抑制导致图像细节信息丢失的问题,提出一种改进相似性度量的BM3D算法。在对相干斑噪声模型分析基础上,采用对数变换使BM3D算法适用于SAR图像去噪,并从相似图像块划分的权重计算方法入手,引入皮尔逊相关系数改进相似性度量准则,从而提升相似图像块匹配准确度。实验选取永城部分地区Sentinel-1A影像数据,并对该算法和其他几种滤波算法进行了对比分析,结果表明本文算法能有效去除斑点噪声,同时保留图像边缘细节信息。Aiming at the problem that BM3 D algorithm suppresses the multiplicative coherent speckle noise of synthetic aperture radar(SAR)images and causes the loss of image detail information,this paper proposes an BM3 D algorithm that improves the similarity measure.Based on the analysis of the coherent speckle noise model,the logarithmic transformation is used to make the BM3 D algorithm suitable for SAR image denoising.And starting from the weight calculation method of the division of similar image blocks,the Pearson correlation coefficient is introduced to improve the similarity measurement criteria.Thereby improving the matching accuracy of similar image blocks.The experiment selects Sentinel-1 A image data in some parts of Yongcheng,and compares and analyzes the algorithm in this paper with several other filtering algorithms.The results show that the algorithm in this paper can effectively remove speckle noise while retaining the image edge detail information.

关 键 词：合成孔径雷达 相干斑噪声 BM3D 相似性度量 

分 类 号：P237[天文地球—摄影测量与遥感]