基于粗糙集理论的SAR图像speckle滤波方法

A method for speckle intelligent filtering of SAR image based on rough sets theory

针对合成孔径雷达(SAR)采用主动式相干波的成像方式,在从原始信号到图像的重建过程中存在斑点噪声的干扰的问题,根据粗糙集理论的条件属性把SAR图像像素分成三类:均匀区域类、非均匀区域类和包含分离点目标类.对不同的像素类采用不同的滤波方式,然后合并三类子图像得到SAR图像的speckle滤波的图像.该方法既可除去图像中的斑点噪声,又可保留图像中的细节特征,有利于对图像的后续处理.

Synthetic aperture radar （SAR） adopted an active imaging way of coherent waves, so an immanent problem of the disturbance of the speckle is posed in rebuilding image from original signals. According to condition attribute based on local statistical model in rough sets theory this paper classified the pixels of the SAR image into three kinds： homogeneity region, inhomogeneity region and insularity pixels point of object. The different filtering ways were used to deal with the different kinds of pixels, and three subimages are merged into a new SAR image with the speckle having been filtered. The method is proved to be practical and efficient by the experiments. It could not only filter speckle noise but also remain fine characteristic,and it is useful for image process.