基于极化SAR相干斑模型的非局部均值相干斑抑制

Non-local Means Speckle Filtering Based on Polarimetric SAR Speckle Model

随着高分辨率和极化合成孔径雷达(Polarimetric Synthetic Aperture Radar,Pol-SAR)技术的发展,Pol-SAR应用前景越来越广阔,但是Pol-SAR数据中相干斑噪声的存在,严重影响了后续目标分类、目标检测和识别跟踪等应用。针对Pol-SAR相干斑去噪问题,提出了基于Pol-SAR相干斑模型的形状自适应非局部均值滤波算法。该算法对于极化协方差矩阵C的对角线元素和非对角线元素分别进行处理,对对角线元素按照乘性噪声模型处理,计算复相关系数,再对非对角线元素进行处理。实验结果表明,所提算法在极化SAR相干斑抑制上具有良好的效果。

With the development of high-resolution and polarimetric synthetic aperture radar(Pol-SAR)technology,the application prospect of Pol-SAR is getting wider and wider,but the existence of speckle noise in Pol-SAR data seriously affects the subsequent applications such as target classification,target detection and target recognition.Aiming at the denoising problem of Pol-SAR speckle,a shape-adaptive non-local means filtering algorithm based on Pol-SAR speckle model is proposed.The algorithm processes the diagonal elements and off-diagonal elements of the polarization covariance matrix C respectively.The diagonal elements are processed according to the multiplicative noise model,the complex correlation coefficient is calculated,and then the off-diagonal elements are processed.Experiment results show that the proposed algorithm has a good effect on speckle filtering in polarimetric SAR.