文献类型：期刊文献

中文题名：灰阶边缘的细化

英文题名：Thinning of Grayscale Edge

作者：陆宗骐[1];梁诚[1]

机构：[1]华东理工大学信息学院电子与通信工程系,上海200237

年份：2003

卷号：24

期号：1

起止页码：107

中文期刊名：小型微型计算机系统

外文期刊名：Journal of Chinese Computer Systems

收录：CSTPCD;;Scopus;北大核心:【北大核心2000】；CSCD:【CSCD2011_2012】；

语种：中文

中文关键词：图象分割;边缘检测;边缘细化

外文关键词：image segmentation; edge detection; edge thinning

摘要：本文提出了一种改进了的灰阶图象 Sobel边缘细化新方法 ,通过对灰阶边缘图多次 Sobel边缘细化结果的叠加使边缘点的数值拔高、变陡 ,再通过灰阶边缘图中各 3 x3邻域内取数值最大的三个点作边缘 ,得到单点宽的边缘 .其中后者也适用于其它边缘检测算子 .此方法处理简单、效果明显 ,为从图象中提取模糊边缘和微弱边缘、提高定位精度 。
An improved new edge thinning method based on Sobel operator is proposed in this paper. The edge pixels are made higher and precipitous by adding up the Sobel thinning results of the grayscale edge image. Then edges of single pixeled width can be obtained by selecting the first three pixels with maximal grayscale among every 3x3 neighboring region in the grayscale edge image, and this method also applies to other edge detection operators. This method provides an effective means for the extraction of blurry or feeble edges. This improves the positioning precision and it is brief and distinct.