The Research On SAR Image Speckle Reduction Based On Partial Differential Equation

Synthetic Aperture Radar (SAR) has an all-weather, day and night, high-resolution ability as a microwave remote sensing technology. It is applied aboard in military and civilian fields. However, SAR images product much speckle inevitably due to the coherent imaging system. The speckle noise greatly lowers the capability from SAR images to distinguish objects of interest and severely affects the sequent SAR image understanding, such as edge detection, SAR image segmentation, and object recognition. So, it is important to suppress speckle for scientific research and practical application.Partial differential equation (PDE), as an important class for image processing, is indispensability to suppress speckle. PDE algorithm is first advanced aiming at optical images. But it is absolutely the same with speckle removal. The difference is just to combine SAR background while constructing diffusion coefficient. This paper researches PDE-based speckle removal approach.SAR images are desired to effectively reduce speckle noise while preserving edges in images. The anisotropic diffusion has been successfully used in speckle reduction. But its performance highly depends upon the measure to discriminate edges based on gradient which is not veracity. In this paper, an improved anisotropic diffusion algorithm with the aid of the ratio test statistic is presented, where the ratio test statistic of constant false alarm rate (CFAR) with multiplicative speckle replaces the traditional gradient-based measure to discriminate edges. The CFAR ratio test statistic can better discriminate edges, particularly in the regions with high scattering intensities. The experimental results using synthetic SAR image and real SAR images show that the proposed algorithm better suppresses speckle and preserves edges than the existing anisotropic diffusion algorithm does.