带钢缺陷图像的自动阈值分割研究  

The Research on Automatic Thresholding of Defect Segmentation for Steel Images

作　　者：陈黎[1] 黄心汉[1] 王敏[1] 何永辉[2] 龚世强 

机构地区：[1]华中科技大学控制科学与工程系,武汉430074 [2]宝钢集团公司技术中心,上海201900

出　　处：《计算机工程与应用》2002年第7期244-246,共3页Computer Engineering and Applications

摘　　要：文章根据带钢缺陷图像的特点,提出了基于灰度统计特性的图像边缘检测方法,结合纹理检测、聚类分析,实现了对带钢缺陷图像的自动阈值分割。实验结果表明,该方法具有在精度和速度方面的优势,适合对带钢表面图像的缺陷进行实时检测。According to the properties of steel images,this paper presents an automatic thresholding method of edge detection based on the properties of gray statistics and combined with texture detection and clustering analysis,automatic defect segmentation of steel images is successfully implemented.The experiment result shows that this method possesses the advantage precision and speed and is useful for the real-time defect segmentation for steel surface image.

关 键 词：图像分割 阈值 边缘检测 聚类分析 带钢缺陷 钢板轧制 

分 类 号：TG316.192[金属学及工艺—金属压力加工]