Abstract: Image segmentation is one of the fundamental and key techniques in the context of SAR image processing, which can affect the performance of SAR image automatic interpretation greatly. SAR image segmentation has been an open problem as the results will be affected greatly by the speckle noise. For solving the problems of the traditional Otsu algorithm, including the low segmentation accuracy, CV model’s sensitivity to the initial level set function and low curve evolution efficiency, a novel fused segmentation algorithm is proposed in this paper to improve the segmentation performance. 1D Otsu algorithm is utilized for coarse segmentation to provide local region and initial situation for CV model, which can simultaneously reduce the scene complexity, increase the segmentation efficiency and reduce the sensitivity of CV model to the initial situation. And image edge intensity, which is derived from mean intensity ratio, is utilized to modify the partial differential of traditional CV model instead of the Dirac function, which can make the proposed algorithm fit the characteristic of SAR image better and speed up the convergence rate. Experimental results show that the fused segmentation algorithm has the advantage of real-time and accurate SAR image segmentation without the setup of initial level set function.