摘要
鉴于Sobel算子检测边缘较粗、定位不准确,以及LoG算子具有各向同性的特点且对边缘方向性信息检测不敏感,提出了Sobel算子与LoG算子相结合的边缘检测与细化方法。首先用水平、垂直、两个斜对角4个方向模板改进原Sobel算子两个方向模板,并用改进的Sobel算子对原图像进行边缘检测,得到粗边缘图像;然后使用LoG算子检测粗边缘图像的边缘;最后将两次边缘检测结果相减,达到边缘细化的目的。实验结果表明,该方法有效地解决了原Sobel算子检测边缘较粗的问题,得到的边缘较细;克服了LoG算子对方向性边缘信息不敏感的缺陷;且运算速度与传统LoG方法基本相当。
The edges detected by Sobel operator are rough and had the poor accuracy and because of its isotropy, traditional Laplacian of Gaussian （LoG） operator is not used in many occasions. So a method for edge detection and thinning combining Sobel operator and LoG operator is put forward. The improved Sobel operator is got with four gradient templates in the horizontal, vertical and two diagonal directions, and the rough edge image is got after edge detection using the improved Sobel operator. And then LoG operator is used to detect th,： rough edge to get a new edge. Finally, edge thinning is got with the difference between the rough edge image and the new edge image. The experimental results show that this method can thin the rough edge detected by Sobel operator effectively, has solved the defect that LOG operator has poor ability to get directional edge information and its speed is roughly equal to LoG operator.
出处
《微型机与应用》
2014年第7期35-37,共3页
Microcomputer & Its Applications
