论文部分内容阅读
该文针对在极化合成孔径雷达(Pol SAR)图像相干斑抑制过程中结构特征和极化散射特性保持的难题,提出一种基于同质像素预选择的非局部均值滤波算法(NLM-HPP)。该算法结合像素统计特性和极化散射机制选择滤波同质像素,并引入结构损失函数提高非局部均值(NLM)算法中像素间相似性度量的准确性,最后用改进的相似性度量对同质像素的协方差矩阵进行加权平均,实现对Pol SAR图像的相干斑抑制。对真实Pol SAR数据进行的实验结果表明,与现有的Refined Lee滤波、基于散射模型的滤波方法和两种非局部均值滤波相比,此方法在有效滤除相干斑点的同时能更好地保持Pol SAR图像的结构信息和极化信息。
In order to solve the problem of structural characteristics and polarization scattering during the speckle reduction of polarimetric synthetic aperture radar (SAR) images, a non-local mean filter algorithm (NLM-HPP) based on homogeneous pixel pre- . This algorithm combines the statistical properties of pixels and the polarization scattering mechanism to select the homogeneous pixels and introduces the structure loss function to improve the accuracy of the similarity measure between pixels in the nonlocal mean (NLM) algorithm. Finally, with the improved similarity measure, Pixel covariance matrix weighted average, to achieve the Pol SAR image speckle suppression. The experimental results on real Pol SAR data show that this method can better keep the coherent speckle while keeping it better than the existing Refined Lee filter, the scattering model based filter and the two non-local mean filters Structure Information and Polarization Information of Pol SAR Image.