在处理合成孔径雷达(SAR)图像中, 精确地估计图像的局部统计特性和尽可能多地滤除相干斑都需要较大的窗口尺寸, 而这与图像边缘和纹理结构的检测需要较小的窗口尺寸相矛盾。一般滤波器都是二者窗口尺寸的均衡, 这必然导致二者都不能取得很好的效果。本文提出的新方法有效的解决了这一矛盾。此方法使处理后的SAR图像既保存了边缘和纹理结构, 避免了以往滤波器中使边缘变模糊的问题, 又有效地滤除了相干斑噪声。

In the processing of Synthetic Aperture Radar (SAR) Images, a large window size, which is often used in estimating the local statistics of the image and filtering the speclke is needed. On the other hand, contrary, detection of edges and textural features often need small sized windows. Normal filters, which are intermediary ofthe two above, don't give better results. This paper presents a new effective algorithm to solve the problem. In this method, the edges and the textural features of the SAR images are presented so the the problem of edge smearing in the normal filtering algorithm is avoid and the speckles are filtered effectively.