Speckle噪声滤波算法的研究与设计

Study and design of a filter algorithm for Speckle noise

针对Speckle噪声的复杂性和真值图像的不可知性,结合奇异值分解和曲面拟合进行了消噪算法的研究与设计.首先对局部图像的奇异值分解结果采用中值滤波进行消噪,其次引入参数修正过程以替代传统的阈值处理方法,最后利用曲面拟合技术恢复局部图像并获得消噪结果.实验表明,与传统的滤波方法相比,从定性和定量方面进行分析,文中算法对Speckle噪声获得了较为满意的结果.

Aiming at the complexity of Speckle noise and the fact that the original image is unknowable, a combination of singular value decomposition（SVD） with surface fitting is used to remove the Speckle noise in image processing. Firstly, SVD is operated on the SVD result of every local image. Secondly, a parameter is introduced to revise the singular values to replace of traditional threshold method. Finally, the grey of centered pixel in local image can be got by surface fitting, and the denoised image can be reconstructed. Experiments show that, comparing to the traditional mean-filter and median-filter, the proposed algorithm can get a more satisfactory result on qualitative and quantitative analysis for Speckle noise.