Abstract: With the processing of the synthetic aperture radar(SAR) imaging,the image can be decomposed into different components.Each component corresponds to different properties,which can be processed by a corresponding algorithm in order to get better performance.A nonlinear diffusion method based image decomposing algorithm is proposed,and the image can be decomposed into the structural component and the textural component which represents the large scale smooth features and the small scale texture features respectively.With the diffusion processing,the diffusion parameter is evaluated,and a parameter,which is the product of the local variance and the mean amplitude,is proposed.The proposed parameter can be used as the criterion of enhancing or weakening the diffusion processing.Finally,the testing results indicate that the diffusion processing has better pertinence for the following target recognizing and image compressing.