文章导读
总览 评价 金剑秋 1, , 梁克维 2,* ( 1、 浙江工商大学计算机与信息工程学院; 2、 浙江大学数学系; ) 摘要: 图像除噪是一个经典的,极富挑战的图像处理问题,而总变差模型方法是当前主要的除噪方法之一。但我们通过实验分析表明,总变差模型方法存在
金剑秋1,, 梁克维2,*
(
1、浙江工商大学计算机与信息工程学院; 2、浙江大学数学系; )
摘要:
图像除噪是一个经典的,极富挑战的图像处理问题,而总变差模型方法是当前主要的除噪方法之一。但我们通过实验分析表明,总变差模型方法存在两个问题:一是计算速度较慢,二是在处理强噪声图像时,会将强噪声点当作图像边缘保留下来,或被过度光滑。针对这两个问题,我们通过在首次迭代中引入光滑算子,和简化总变差模型,得到一种高效的基于WLS(weighted least square)的图像除噪方法。实验表明,我们的方法无论在速度上和除噪效果上较总变差模型方法和基于偏微分方程的滤波方法有明显的改善。
关键词:
除噪算法;偏微分方程;WLS
Jin Jianqiu1,, Liang Kewei 2,*
(
1、College of Computer Science & Information Engineering, Zhejiang Gongshang University; 2、Department of Maethematics, Zhejiang University; )
Abstract:
Total variation (TV) technique has become one of the most popular and successful methodology for image denoising. But our experiments show it can not quickly get the solution of ROF model, and it is involved in a dilemma whether the strong noise is preserved as the edges or the result image is over smooth. In order to overcome these shortages, we propose a novel image denoising algorithm based on WLS (weighted least square). This algorithm, which is effective, simple and fast, is accomplished by a smooth operator in the first iteration and solving a simplification of TV model by iterations. Experimental results demonstrate the efficient of the proposed algorithm.
Tag:
点此返回栏目查看更多>>>参考论文