基于Canny算子的图像边缘检测优化算法  

An Optimal Algorithm of Image Edge Detection Based on Canny

作　　者：徐亮[1] 魏锐[1] 

机构地区：[1]黄淮学院信息工程学院,河南驻马店463000

出　　处：《科技通报》2013年第7期127-131,150,共6页Bulletin of Science and Technology

基　　金：河南省科技攻关项目(122102210510)

摘　　要：针对传统的Canny算子进行边缘检测时易丢失边缘细节的缺陷,提出一种改进的Canny的边缘检测算法。首先用开关型中值滤波器抑制噪声,保持图像的边缘信息,然后采用直方图凹度分析自动选取双阈值,并用此高低阈值检测及连接图像的边缘,从而得到较为理想的图像边缘。实验结果表明,改进后的Canny算子具有很强的阈值自适应能力,能有效检测到图像中的细节梯度,并去除了伪边缘和噪声边缘,有效提高了边缘的检测精度和准确性。The traditional Canny edge detection algorithm exists the defect losing details of the edge,so this paper proposed an improved Canny edge detection algorithm.Firstly,by switching median filter is robust to noise,Keep the edge information of image,Then the automatic selection of threshold histogram concavity analysis,and the level of the threshold detection and edge linking image,in order to get the more ideal image edge.The experimental results show that,the improved Canny operator with threshold after strong adaptive capacity,can effectively detect the details of image gradient,and remove the false edges and noise margin,effectively improve the edge detection precision and accuracy.

关 键 词：图像边缘检测 直方图凹度分析 CANNY算子 阈值 

分 类 号：TP317.4[自动化与计算机技术—计算机软件与理论]