混合噪声的各向异性扩散平滑

来源期刊:中南大学学报(自然科学版)2010年第1期

论文作者:马捷 严家斌 刘贵忠 柳建新

文章页码:231 - 237

关键词:混合噪声;各向异性扩散;飞点;边缘

Key words:mixed noise; anisotropic diffusion; outlies; edge

摘    要:针对混合噪声模型中中值滤波、线性滤波与传统各向异性扩散存在固有的缺陷,通过修改扩散项的计算方式改进扩散模型并提高扩散模型的滤波特性,改进模型具有带通滤波特性,扩散过程稳定并满足极值原理。研究结果表明:中值滤波作为一种经典的非线性滤波方法仅适合于“飞点”噪声的滤波并引起图像的角点、线条损失,线性滤波扩大或使图像边界范围模糊;传统的各向异性扩散能对随机噪声有效地压制,但对“飞点”噪声会产生虚假的边缘甚至反扩散,而改进模型既具有中值滤波的特性又具有传统各向异性扩散的性能,适合于平滑不同强度与形状的混合噪声,并对角点、线条、边缘等细节信息提供足够保护。

Abstract: Based on the fact that there is intrinsic drawback for median filter, linear filter and tradition anisotropic diffusion to mixed noise, a new diffusion model was proposed by modifying the calculation method of diffusion term and improving filter capability of the model. The improved model possesses band-pass filer characteristic, diffuses processing stability and satisfies the max-min principle. The simulation results show that the median filter is only proper for outlies noise and removes the thin line and corners of image. The linear filter extends or blurs the boundary of image. The anisotropic diffusion removes random noise effectively, but it brings about illusive edge and back diffusion for outlies noise. The improved model that possesses both median filter characteristics and anisotropic diffusion capability is proper for various mixed noise and can sufficiently protect the detailed characteristic such as corner, thin line and edge.

基金信息:国家自然科学基金资助项目
湖南省自然科学基金资助项目

有色金属在线官网  |   会议  |   在线投稿  |   购买纸书  |   科技图书馆

中南大学出版社 技术支持 版权声明   电话:0731-88830515 88830516   传真:0731-88710482   Email:administrator@cnnmol.com

互联网出版许可证:(署)网出证(京)字第342号   京ICP备17050991号-6      京公网安备11010802042557号