Abstract: It is difficult to preserve the edge and structure information while suppressing the speckle for Synthetic Aperture Radar（SAR） image despeckling. To resolve this issue, the intensity association method for SAR image despeckling is proposed. Pixels are separated into two parts using intensity association method with global threshold. And then the part with larger number of pixels is selected to estimate the value of the center pixel using Maximum Likelihood（ML） estimation. Furthermore, an evaluation method based on statistical spectrum similarity is also proposed for SAR despeckling evaluation. Simulations and real SAR images are processed to demonstrate the validity of proposed method.

Key words: Synthetic Aperture Radar（SAR）image, speckle reduction, intensity association, statistical spectrum similarity

摘要： 在合成孔径雷达（Synthetic Aperture Radar，SAR）图像相干斑抑制过程中，如何在去斑的同时保持图像的边缘和结构信息是一个难点问题。针对此问题提出了SAR图像幅度关联去斑算法，该算法基于幅度关联，通过全局幅度关联阈值将滑窗内像素分成两个部分，选取像素数目较多的一部分使用最大似然估计方法估计出中心像素的真实值。提出了一种基于频谱统计相似度的评估方法，并利用该评估方法及一些经典的去斑性能评估方法，分别通过仿真数据和实测数据处理结果验证了所提去斑算法的有效性。

关键词: 合成孔径雷达（SAR）图像, 相干斑抑制, 幅度关联, 频谱统计相似度