基于视觉梯度的图像非线性扩散滤波  

Noise removal model with anisotropic diffusion based on visual gradient

作　　者：李世飞[1] 王平[1] 沈振康[1] 

机构地区：[1]国防科技大学ATR国家重点实验室,长沙410073

出　　处：《计算机工程与应用》2010年第2期167-169,共3页Computer Engineering and Applications

基　　金：国家部委预研基金资助项目~~

摘　　要：各项异性非线性扩散滤波方法自诞生以来,许多研究者致力于该方法的改进和优化,并取得了很大的进步。然而,这些改进模型都很少关注到人类视觉系统信息的利用。把视觉系统信息融合到梯度信息中,定义了视觉梯度这一全新的概念。在此基础上,用视觉梯度替代经典的各项异性扩散模型中的图像梯度,得到了基于视觉梯度的各向异性非线性扩散滤波模型。最后,数值实验结果说明了该模型的效果。In recent years triggered by the anisotropic diffusion method considerable amount of researchers have been devoted to this and related methods and achieved a series of important development.However,HVS（Human Visual System） which perceives and interpretes images has been paid little attention to in all these models.This paper defines visual gradient which is looked as a generalization of image gradient.After that substitutes the visual gradient for image gradient in the classical anisotropic diffusion model’s diffusion function to keep to some extent consistent with human visual system for the first time.Numerical results show the proposed method’s performance.

关 键 词：扩散滤波 视觉梯度 图像 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]