基于快速非局部均值滤波的SAR图像相干斑抑制算法  

Despeckling Algorithm for SAR Image Based on Fast Non-local Mean Filtering

作　　者：李浠铭 甘新胜[1] 程健庆[1] LI Xi-ming;GAN Xin-sheng;CHENG Jian-qing(Jiangsu Automation Research Institute, Lianyungang 222061, China)

机构地区：[1]江苏自动化研究所,江苏连云港222061

出　　处：《指挥控制与仿真》2020年第6期63-68,共6页Command Control & Simulation

摘　　要：针对SAR图像的相干斑噪声抑制问题,提出了一种基于快速非局部均值滤波(Fast None-local Means,FNLM)的SAR图像去噪算法,该方法首先将乘性相干斑噪声模型进行变换,再利用积分图像计算欧氏距离以降低时间复杂度,最后将Turkey权函数与高斯权函数相结合成一种自适应分段的权函数用于权值分配。实验结果表明,改进FNLM能够在保证快速的同时有效地抑制相干斑噪声。Aiming at the problem of multiplicative coherent speckle noise suppression in SAR images,a SAR image denoising algorithm based on Fast Non-local Mean Filtering(FNLM)is proposed.This method firstly transforms the multiplicative coherent speckle noise model.Then,the integrated image is used to calculate the Euclidean distance to reduce the time complexity.Finally,the Turkey weight function and the Gaussian weight function are combined into an adaptive segmented weight function for weight distribution.Experimental results show that the improved FNLM can effectively suppress coherent speckle noise while ensuring low time complexity.

关 键 词：合成孔径雷达图像 非局部均值 对数变换 权函数 相干斑抑制 

分 类 号：E919[军事] TP274[自动化与计算机技术—检测技术与自动化装置]