基于视觉特性的各向异性扩散滤波器改进方法  

An Improved Method of Anisotropic Diffusion Filter Based on Visual Characteristics

作　　者：龚昌来[1] 罗聪[1] GONG Chang-lai;LUO Cong(School of Electronic & Information Engineering, Jiaying University, Meizhou 514015, China)

机构地区：[1]嘉应学院电子信息工程学院

出　　处：《嘉应学院学报》2019年第3期29-33,共5页Journal of Jiaying University

基　　金：广东省自然科学基金资助项目(S2012010010368)

摘　　要：比较了几种现有基于视觉特性各向异性扩散方程中扩散系数的优缺点,提出了一种改进的扩散系数.以局部平均梯度替代原扩散系数中局部方差,减少了运算量;用图像的最大值灰度值替代原扩散系数中的常数1,使扩散系数趋于零的速度变慢,提高了收敛的速度.实验结果表明,改进的扩散系数在滤波去噪效果和运算时间方面优于其它同类扩散系数.Based on the comparisons of the advantages and disadvantages of several existing visual characteristics of anisotropic diffusion coefficient, an improved diffusion coefficient is proposed. Original local variance of diffusion coefficient has been replaced by local average grads, which reduces the amount of computation. The constant 1 of the original diffusion coefficient has been replaced by the maximum gray value, which makes the speed of the diffusion coefficient tending to zero slowly, while speeds up the convergence. The experimental results show that, the improved diffusion coefficient is superior to other similar diffusion coefficient in the smooth denoising effect and the calculation time.

关 键 词：各向异性扩散方程 扩散系数 视觉掩蔽效应 局部平均梯度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]