基于视觉特性的Riemann-Liouville分数阶图像增强
A Riemann-Liouville Fractional Differential Image Enhancement Algorithm Based on Human Visual Characteristics
作者：陈庆利(乐山师范学院 计算机科学学院;四川大学 计算机学院)；黄果(乐山师范学院 计算机科学学院)；孙锐(乐山师范学院 计算机科学学院)；舒勇(乐山师范学院 计算机科学学院)；蒲亦非(四川大学 计算机学院)；周激流(四川大学 计算机学院)
Author：Chen Qingli(Dept. of Computer Sci.,Leshan Normal Univ.;School of Computer Sci., Sichuan Univ.)；Huang Guo(Dept. of Computer Sci.,Leshan Normal Univ.)；Sun Rui(Dept. of Computer Sci.,Leshan Normal Univ.)；Shu Yong(Dept. of Computer Sci.,Leshan Normal Univ.)；Pu Yifei(School of Computer Sci., Sichuan Univ.)；Zhou Jiliu(School of Computer Sci., Sichuan Univ.)
收稿日期：2011-05-24 年卷（期）页码：2012,44(1):99-105
期刊名称：工程科学与技术
Journal Name：Advanced Engineering Sciences
关键字：图像增强;分数阶微分;视觉特性;感受野;分数阶微分模板
Key words：image enhancement;fractional difference;visual characteristic;receptive field;fractional differential mask
基金项目：国家自然科学基金资助项目(60972131);四川省教育厅科研基金资助项目（11ZB132）
中文摘要
为了解决Grünwald-Letnikov分数阶微分算法对彩色图像增强容易产生色彩失真以及传统方法增强效果不明显的问题，在分数阶微分视觉模型的基础上提出了一种基于Riemann-Liouville(R-L)分数阶微分的数字图像增强算法。首先讨论了数字图像的分数阶微分视觉模型；接着在R-L分数阶微分方程的基础上构造了8个方向上的分数阶微分增强模板；并讨论了这些微分增强模板的数值运算规则；并在HSI色彩空间对I分量进行分数阶微分实现彩色图像的增强处理。实验表明本文方法具有非线性特性，对图像增强效果明显，且增强
英文摘要
In order to overcome the defects that enhancement effects of classical approaches are not obvious and the Grünwald-Letnikov(G-L) fractional differential operator could make color image distortion, a fractional differential image enhancement algorithm based on fractional-order receptive field model for gray and color image enhancment was presented.First, fractional-order receptive field of digital image was discussed.Next, a fractional-order differential convolution mask for image enhancement was constructed according to R-L fractional partial differential equation and its numerical implementation on eight directions was discussed in detail. This algorithm was applied in HSI color space and only the Intensity(I) component was processed for color image enhancement.The experiments showed that the proposed algorithm is non-linear and it can obviously enhance gray and color images.The enhanced images have no color distortion.
【关闭】
