| 基于非局部均值滤波的SAR图像抑噪 |
| |
| 引用本文： | 张文哲,潘军,蒋立军,仲伟敬,张雪峰,范博文,卞宇涛.基于非局部均值滤波的SAR图像抑噪[J].世界地质,2020,39(2):462-470. |
| |
| 作者姓名： | 张文哲 潘军 蒋立军 仲伟敬 张雪峰 范博文 卞宇涛 |
| |
| 作者单位： | 吉林大学 地球探测科学与技术学院, 长春130026;中国人民解放军63771部队, 西安714000 |
| |
| 基金项目： | 高等学校博士学科点专项科研基金(新教师类) |
| |
| 摘 要： | 针对非局部均值(non-local means,NL-means)滤波算法在对合成孔径雷达(synthetic aperture radar,SAR)图像进行滤波抑噪处理时存在使图像过平滑而丢失纹理细节的问题,笔者提出一种改进非局部均值滤波的SAR图像相干斑噪声抑制算法。该算法将图像梯度加入非局部均值滤波算法,改进相似块的权值计算,使滤波算法在有效抑制相干斑噪声的同时能够更好地保留图像的纹理结构信息而不使图像过平滑。通过在真实SAR图像上的实验表明,与Refined Lee、Gamma MAP、双边(BF)及NL-means滤波算法相比,本文算法可有效抑制相干斑噪声、平滑图像,同时能够更大程度地保留纹理结构信息。
|
| 关 键 词： | 合成孔径雷达图像 抑噪 非局部均值 梯度 |
SAR image despeckling based on non-local means filtering |
| |
| ZHANG Wen-zhe,PAN Jun,JIANG Li-jun,ZHONG Wei-jing,ZHANG Xue-feng,FAN Bo-wen,BIAN Yu-tao.SAR image despeckling based on non-local means filtering[J].World Geology,2020,39(2):462-470. |
| |
| Authors: | ZHANG Wen-zhe PAN Jun JIANG Li-jun ZHONG Wei-jing ZHANG Xue-feng FAN Bo-wen BIAN Yu-tao |
| |
| Institution: | (College of Geo-exploration Science and Technology,Jilin University,Changchun 130026,China;63771 PLA Troops,Xi an 714000,China) |
| |
| Abstract: | To solve the problem of losing texture detail due to excessive smoothing in synthetic aperture radar(SAR)images despeckling with non-local means(NL-means)filter algorithm,the authors proposes a improved non-local mean filter algorithm for SAR image despeckling.It adds the gradient of image into the non-local mean filter algorithm to optimize the weight calculation of similar blocks,so that the filter algorithm can preserve texture and structure information of the image better while despeckling effectively,instead of over smoothing the images.The experimental results on real SAR images show that compared with the Refined Lee,Gamma MAP,Bilateral filter and NL-means filter algorithms,the proposed algorithm can retain the texture and structure information in a greater extent while despeckling and smooth images effectively. |
| |
| Keywords: | SAR image despeckling non-local means gradient |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
