In order to effectively preserve SAR image edges while filtering,the authors propose an anisotropic diffusion filtering algorithm for speckle noise based on speckle reduction anisotropic diffusion (SRAD) model.On one hand,it can be proved that the diffusion coefficient in the proposed algorithm theoretically satisfies the conditions for the design of diffusion coefficients presented by Charbonnier,et al;on the other hand,the proposed algorithm can achieve a series of different filtered results,through the simple adjustment of two parameters.One parameter is a threshold of the accumulative histogram in the edge image,and the other one is a threshold of the relative signal to noise ratio.And the algorithm is applicable in different applications,including mapping and applications in which fine details and high resolution are required.The experimental results show that the proposed algorithm has capabilities of efficiently filtering speckle noise and preserving edge and texture features.Moreover,it outperforms the four traditional filtering methods in speckle smoothing,edge and texture preservation and visual assessment.
ZHANG Liang-pei, WANG Yi, LI Ping-xiang.
The Filtering Algorithm for SAR Image Speckle Noise Based on Anisotropic Diffusion[J]. Acta Electronica Sinica, 2006, 34(12): 2250-2254.