基于Sigma方法的SAR图像斑点噪声滤波器  

An Improved SAR Image Speckle Noise Filter Based on Sigma Method

作　　者：苑云[1,2] 李壮[1,2] 朱宪伟[1,2] 

机构地区：[1]国防科学技术大学航天与材料工程学院,长沙410073 [2]图像测量与视觉导航湖南省重点实验室,长沙410073

出　　处：《航天返回与遥感》2011年第4期37-44,共8页Spacecraft Recovery & Remote Sensing

基　　金：国家自然科学基金(60904084);863项目支持(2007AA12Z121)

摘　　要：文章基于传统Sigma滤波器,提出了一种新的SAR图像滤波方法。首先,分析指出传统Sigma滤波器处理复合噪声和保持弱边缘的能力较弱的缺点。然后,改进Sigma滤波器的δ-条件,使之更加适应复合噪声分布条件下的SAR图像滤波。提出了灰度中心偏移的Sigma滤波(GCOSF),增强Sigma滤波器保持弱边缘的能力。在GCOSF中,采用局部方差系数检测边缘区域,对边缘区和非边缘区分别使用不同的滤波策略。最后,采用机载SAR图像测试文章所提出的滤波方法。实验结果表明,该方法能够有效滤除复合噪声,同时较好地保持弱边缘细节信息,明显优于传统的SAR图像滤波方法,因此具有较高的实用价值。This paper proposed an improved SAR image speckle denoising method based on Sigma filter.First,it introduced the traditional Sigma filter and showed its weakness in removing complex noise and keeping weak edge.Second,the paper modified the δ-condition in Sigma filter to accommodate it to complex noise.Then,to maintain weak edge,it put forward a new method named gray center offset Sigma filter（GCOSF）.In GCOSF,local variance coefficient was utilized to detect regions which contain edges.Different filter strategies were applied to smooth region and marginal region.In the end,the results show that the GCOSF performed better than traditional Sigma filter and several other filters both in noise removing and edge keeping.

关 键 词：图像处理 斑点噪声 复合噪声 西格玛滤波器 弱边缘信息 合成孔径雷达 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]