All direction auto-adaptive dynamic window filter for noise suppression in SLC SAR image

To smooth coherent speckle noise and preserve edge information in SLC SAR images as precise as possible, a new algorithm called all direction auto-adaptive dynamic window filtering method based on coherent speckle statistic characteristic and analysis of spatial filtering algorithms for SAR image, is developed in this paper. For every processing pixel the filtering window is divided into mutually exclusive all direction sub-windows according to the complexity of image texture and edge existence. Local relative standard deviation is used to determine whether the local filtering window area is homogeneous. Then a Kuan filter and a 3×3 directive operator are incorporated to process the SAR image. The proposed method can auto-adaptively modulate its filtering window size and selection of filtering pixels. The developed method is applied to a single-look ERS SAR image. Experimental results show that the performance of the method is satisfactory in both speckle suppression and preservation of image details.