一种基于图像融合的含噪图像边缘检测方法  

An edge detection method based on image fusion in a noisy image

作　　者：乔闹生[1,2] 邹北骥[2] 邓磊[1] 曾友兵[1] 邹剑臣[1] 

机构地区：[1]湖南文理学院物理与电子科学学院,湖南常德415000 [2]中南大学信息科学与工程学院,湖南长沙410083

出　　处：《光电子．激光》2012年第11期2215-2220,共6页Journal of Optoelectronics·Laser

基　　金：国家自然科学基金(61144006;61173122;60970098;60803024);国家自然科学基金重大研究计划(90715043);中央高校基本科研业务费专项资金(201021200062);中南大学博士后科研启动基金(1332/74341016030);湖南文理学院教授(博士)工作室指导学生创新创业资助项目

摘　　要：针对含噪声图像在边缘定位、噪声抑止、弱边缘保留及边缘细节的视觉感知等方面难以兼顾的具体情况,提出了一种基于图像融合的含噪图像边缘检测方法。首先,讨论了基于小波变换模极大值的图像边缘检测法;结合数学形态学基本运算与信息熵理论,提出了改进的灰值数学形态学图像边缘检测法,并分析了其基本原理。然后,结合两种方法的优点,采用叠加运算进行图像融合最终检测到的图像边缘连续完整。最后,与传统边缘检测法、小波变换模极大值法、改进的灰值数学形态学法进行了实验对比,结果表明,本文算法优质系数最大,所花时间最少。In a noisy image,it is difficult to comprehensively consider the aspects of edge orientation, noise restraint, weak edge reservation and vision sense of edge detail, so a method of edge detection based on image fusion is proposed in this paper. Firstly, the method of image edge detection based on wavelet transform modulus maxima is discussed. Combining the basic operation of mathematical morphology with information entropy theory, an improved edge detection method with gray value mathematical morpholo- gy is proposed, and its basic principle is analyzed. Secondly,the final better image edge is obtained by a- dopting superposition operation to fuse the image edges obtained by the both methods. Finally, the exper- imental comparison of our method with traditional edge detection methods, wavelet transform modulus maxima method and improved gray value mathematical morphology method has been implemented and the results verify the feasibility and validity of our method.

关 键 词：边缘检测 图像融合 小波变换模极大值 灰值数学形态学 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]