分数阶微分算子增强图像边缘和纹理  

Fractional differential operator for enhancing edge and texture of image

作　　者：周昌雄[1] 陶文林[1] 尚丽[1] 颜廷秦[1,2] 马国军[3] 

机构地区：[1]苏州市职业大学电子信息工程系,江苏苏州215104 [2]苏州市数字化设计与制造技术重点实验室,江苏苏州215104 [3]江苏科技大学电子信息学院,江苏镇江212003

出　　处：《激光与红外》2013年第1期109-113,共5页Laser & Infrared

基　　金：国家自然科学基金项目(No.60970058);江苏省自然科学基金项目(No.BK2009131);苏州市职业大学创新团队建设项目(No.3100125);苏州市基础设施建设计划项目(No.SZS201009);江苏省青蓝工程项目资助

摘　　要：为了提高增强图像边缘和纹理的能力,分析了分数阶微分算子频域特性,大幅提升信号高频成分同时增强信号的中频成分以及非线性保留信号的甚低频成分,据此可增强图像的边缘信息和纹理细节信息。从分数阶微分定义出发,构建了近似微分算子模板及线性滤波算法。考虑到边缘和纹理具有方向性、空间邻域内像素的关联性以及相邻像素灰度值的相近性,提出在3×3模板窗口内,取中心点四个方向最大值为增强值。实验结果表明,本文提出的分数阶微分算子能明显地增强图像的边缘和纹理信息,增强后图像清晰度提高,视觉效果明显。In order to improve the ability of enhancing image＇s edge and texture, the fractional differential operator for frequent domain characteristics is proposed and analyzed. It shows that the fractional differential can greatly increase high frequency component, reinforce medium frequency component and nonlinearly preserve low frequency component of the signal, hence they could be used for edge and texture enhancement. Starting from the definition of the fractional differential, an approximate fractional order differential module from which a linear filter is adopted is constructed. Im age edge and texture have direction, and there is spatial correlation between the neighbored pixels, so there is the adja cent gray value. In this paper, an algorithm is proposed to improve the ability of fractional differential operator. Firstly, 3 x 3 template is divided into four subtemplates with different directions around the detecting pixel, and then the four weight sum values for the four subtemplates are obtained. Consequently, an enhancing value which is maximum value of the four weight sum values is gained. The experiment results of edge enhancement show that the improve algorithm can enhance edge and texture information effectively more than 5 x 5 template operators.

关 键 词：分数阶微分 图像增强 边缘和纹理 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]