中文摘要：

本文阐述了基于散度的和基于二阶导数的图像去噪算法之间的关系,提出了新的基于二阶导数框架的图像去噪算法,给出了切向扩散系数以及法向扩散系数.实验结果表明：当法向扩散系数为递减函数,在边缘区域该系数的值较小,有效地保留了法向方向的边缘,在平滑区域扩散该系数的值较大,起去噪作用;切向扩散系数则维持在较大的常量对切向方向噪声起较强的去噪作用;在此情况下基于二阶导数的算法能够取得较好的图像去噪效果.

英文摘要：

The relationship between divergence-based and second derivative-based algorithm for image denoising is described and then a new second derivative-based algorithm is put forward.Moreover,tangle diffusion coefficient and normal diffusion coefficient are proposed.Experimental results show that when normal diffusion coefficient is decreasing function and the larger gradient the smaller coefficient,the edges in normal direction are preserved;when the smaller gradient the larger coefficient,the noises in normal direction are smoothed.Tangle diffusion coefficient retains large constant value and always smoothes noises in tangle direction.Therefore the good results are obtained with the algorithm.