基于Prewitt算子的TV图像去噪算法  

A Prewitt-based Algorithm on TV Image Denoising

作　　者：郑秀清[1,2] 何坤[1] 张健[1] 

机构地区：[1]四川大学计算机科学学院,四川成都6100065 [2]四川师范大学计算机科学学院,四川成都610066

出　　处：《四川师范大学学报（自然科学版）》2013年第3期456-462,共7页Journal of Sichuan Normal University（Natural Science）

基　　金：四川省教育厅自然科学基金(09ZB067)资助项目

摘　　要：分析了传统TV算法抑噪不充分的原因,为了弥补其不足结合图像空间和像素梯度提出了一种改进算法.该算法首先运用Prewitt边缘检测算子分析图像空间梯度,并对空间梯度进行抑制,一定程度上克服了传统TV算法对平坦区抑噪不充分,甚至出现虚假边缘和阶梯效应的缺陷;再结合像素梯度分析了TV去噪的迭代函数完成保边去噪.与传统算法相比,该文算法残余噪声较小且边缘保护较好,提高了图像的PSNR和视觉效果.但对图像较小纹理误作为噪声而去除.In this paper, an improved algorithm on TV image denoising is proposed. The method utilizes the prewitt operator to de- tect the spatial gradient, and then suppresses the spatial gradient by theshholding, which solves the TV defaults such as unsufficient noise suppression, the false edges and ladder effects. And then combining with pixel gradiem, the method removes the noises of the images and preserves the edges of the images effectively. Compared with traditional image denoising algorithm, experimental results show that the algorithm has better edge-protection and less residual noise on PSNR and visual effect. However, small textures are re- moved which are regarded as noises.

关 键 词：图像去噪 TV算法 空间梯度 像素梯度 Prewitt 

分 类 号：TP394.1[自动化与计算机技术—计算机应用技术]