去除乘性噪声的积分微分方程模型  

Model based on the integro-differential equation for multiplicative noise removal

作　　者：白键[1] 冯象初[1] 

机构地区：[1]西安电子科技大学理学院,陕西西安710071

出　　处：《西安电子科技大学学报》2013年第3期132-138,共7页Journal of Xidian University

基　　金：博士点新教师基金资助项目(20100203120010);国家自然科学基金资助项目(61105011)

摘　　要：恢复含乘性噪声的图像是当前图像处理的重要研究课题.提出基于积分微分方程的乘性噪声去除模型.新模型中假定乘性噪声服从伽马分布.讨论了经典的去除乘性噪声的AA模型,在此基础上通过在逆尺度空间积分"尺度"图像,从而得到了新的积分微分方程.这种新的积分微分方程含有一个单调增加的尺度函数.通过选取适当的尺度函数,该方程可以有效地去除乘性噪声.Multiplicative noise removal is an important research topic on image processing. This paper presents a novel integro-differential equation approach for removing multiplicative noise. Under the assumption that multiplicative noise follows a Gamma distribution, we firstly discuss the classical AA model, and then in order to arrive at the novel integro-differential equation, integrate in inverse scale space a succession of refined ＇slices＇ of the image. The novel integro-differential equation includes a monotone increasing scaling function. By choosing an adaptive scaling function, this equation can remove multiplicative noise efficiently. Finally, the experimental results demonstrate the better performance of the proposed model.

关 键 词：图像去噪 总变差最小 乘性噪声 积分微分方程 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]