基于矩阵对数累积量的极化合成孔径雷达数据去噪方法

发布时间：2018-12-19 13:48

【摘要】：噪声抑制是极化合成孔径雷达(PolSAR)数据处理中的重要步骤,首次将矩阵的对数累积量应用于PolSAR去噪当中,由数据点的子视数据集计算滤波系数并结合非局部方法的思想实现了PolSAR的斑点噪声抑制。所用方法基于非局部策略选取同质区域,然后利用相干矩阵对数累积量计算得到的滤波系数进行加权滤波。该方法的优点在于同质区域选取和滤波系数的计算均是针对相干矩阵进行的,相对于仅使用主对角线上的元素或SPAN图像,其对极化数据的处理更具合理性。与其他PolSAR去噪算法的对比实验结果表明,该方法在有效平滑同质区域的同时,能够更好地保留地物结构、细节以及纹理信息,并能更好地保持数据的极化相关性。
[Abstract]:Noise suppression is an important step in (PolSAR) data processing of polarimetric synthetic aperture radar. The logarithmic cumulant of matrix is applied to PolSAR denoising for the first time. The speckle noise suppression of PolSAR is realized by calculating the filter coefficient from the sub-view dataset of the data points and combining the idea of nonlocal method. The method is based on the nonlocal strategy to select the homogeneous region and then use the logarithmic cumulant of the coherent matrix to calculate the filter coefficient for weighted filtering. The advantage of this method is that the selection of homogeneous region and the calculation of filtering coefficients are based on the coherent matrix, and the processing of polarization data is more reasonable than using only elements or SPAN images on the main diagonal line. Compared with other PolSAR denoising algorithms, the experimental results show that the proposed method not only smoothes the homogeneous region effectively, but also preserves the structure, details and texture information of the ground objects, as well as the polarimetric correlation of the data.
【作者单位】：西安电子科技大学智能感知与图像理解教育部重点实验室智能感知与计算国际联合研究中心;
【基金】：国家重点基础研究发展计划项目(2013CB329402) 国家自然科学基金项目(61271302) 高等学校学科创新引智计划项目(B07048) 教育部长江学者和创新团队发展计划项目(IRT1170)
【分类号】：TN957.52

【相似文献】