基于二维Otsu准则的自适应分数阶微分的图像增强算法  

An algorithm for image enhancement based on adaptive fractional differential using two-dimensional Otsu standard

作　　者：李博[1] 谢巍[1] 

机构地区：[1]华南理工大学自动化科学与工程学院,广东广州510640

出　　处：《控制理论与应用》2015年第6期794-800,共7页Control Theory & Applications

基　　金：国家自然科学基金面上项目(61174098);广东省自然科学基金项目(S2013010012537);广东省省部产学研合作专项项目(2013B090600062)资助~~

摘　　要：针对传统的分数阶微分应用于图像增强中会在强化图像边缘的时候忽视了图像的纹理,或者在保留更多图像纹理的同时弱化了图像边缘等不足,本文提出一种可以根据像素点的动态梯度来自适应调整分数阶微分阶次的图像增强新方法.该方法引入改进的二维Otsu准则,并结合图像的区域特征构造出自适应分数阶微分函数,进而求出与每一个像素点相对应的分数阶微分阶次.最后,实验结果表明该方法比较现有的方法可以更好的提取和增强图像边缘的同时,保留图像弱纹理和平滑区域,从而达到更佳的图像增强效果.Because traditional fractional differential methods ignore textures of image when enhancing the edges,or weaken the edges when preserving the textures in the process of image enhancement,we put forward a new image enhancement method which adjusts the fractional differential order according to the dynamic gradient of the pixels.The improved two-dimensional Otsu algorithm is employed in designing the adaptive image enhancement operator.In addition,the optimal differential order of every pixel is determined by the adaptive fractional differential function constructed on the basis of the area-feature of image.Consequently,the image is adaptively enhanced by an adaptive fractional differential mask.Experimental results show that this method has better image enhancement than other current methods by making the edges clearer and the weak textures richer.

关 键 词：分数阶微分 自适应算法 图像增强 梯度方法 OTSU 掩模 阶次 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]