基于AOS非线性扩散的SAR图像去噪研究

A study of SAR images denoising based on AOS nonlinear diffusion

针对合成孔径雷达 (SAR)图像的斑点噪声 ,介绍了一种基于非线性各向异性扩散的去噪方法 ,该方法经过加性算子分离 (AOS)方案离散可以保证其扩散迭代过程中滤波结果的绝对稳定 ,并且利用其在消除噪声的同时锐化边界的特点 ,将之引入SAR图像的斑点噪声抑制问题当中。通过对一幅SAR图像的滤波处理 ,以及若干衡量滤波算法效果的评价指标 ,将其与传统的自适应局域统计滤波方法进行分析比较 ,并得出相关结论 ,从而证实了该研究提出的AOS非线性扩散滤波法 (AOSND法 )

Based on nonlinear anisotropic diffusion, a new denoising method is proposed for speckle reduction in the SAR image. A numerical additive operator splitting (AOS) scheme is used to efficiently ensure the filtering stability. A good performance is demonstrated by an example of SAR image in both reducing the noise and preserving important features at the same time. The result is also compared with traditional adaptive filtering algorithm based on local statistics. It shows good feasibility and validity of the proposed algorithm.