一种改进的Canny边缘检测算法  

An Improved Canny Edge Detection Algorithm

作　　者：王兰[1] 吴谨[1] 

机构地区：[1]武汉科技大学信息科学与工程学院,湖北武汉430081

出　　处：《微计算机信息》2010年第2期198-199,202,共3页Control & Automation

基　　金：基金申请人:吴谨;项目名称:无源毫米波成像探测机理及关键技术研究;基金颁发部门:湖北省教育厅(D200711003)

摘　　要：边缘检测是计算机视觉中的研究热点,Canny算法是其中较为典型的算法。本文分析了传统Canny边缘检测算法的原理,针对传统Canny算法在非极大值抑制和高低阈值选取部分做了一定的改进。将像素的八邻域分为4个象限结合线性插值提高了边缘检测的精度;利用梯度直方图选取高低阈值,提高了算法的自适应性。实验表明该方法提高了Canny算子的边缘定位精度,也使算法具有了一定的自适应性。In computer vision, edge detection is a hot research area, in which Canny algorithm is a typical algorithm. This paper analyses the theory of the traditional Canny edge detection algorithm and does some improvements on the parts of non–maximal suppression and high or low threshold selecting. The way dividing the eight neighboring region of pixel to four-quadrant and combining with linear interpolation increases the precision of edge detection. Utilizing the gradient histogram to select the high or low threshold increase the adaptive of algorithm. The experiment results indicate that the improved algorithm enhances the Canny operator on the verge of positioning accuracy and the algorithm has also a certain degree of adaptability.

关 键 词：CANNY算子 边缘检测 非极大值抑制 梯度直方图 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]