An SAR image segmentation method is proposed,which based on the generation characteristics of SAR.Firstly,according to statistical characterisation of SAR images,a new region growing criterion is defined to get the initial image segmentation.Then weighted undirection graph is constructed with the vertexes and the interregional similarity as the edge weight.Finally,the the minimum cut method using the global information is used to combining the regions and get the final segmentation result.The proposed method has the advantage of using both of the local information and global information.The experiment results showed the effectiveness of the approach.