摘要
针对曲面玻璃上划痕缺陷成像难、提取难等问题,提出一种基于机器视觉的划痕缺陷检测方法。首先,通过数学建模分析,推导出曲面曲率、相机位置、打光角度三者之间对应关系。然后,在该关系指导下,获取不同打光角度的缺陷图像,并提出一种基于双阈值分割的缺陷提取算法。最后,实验结果表明,建模推导出打光角度下获取的划痕缺陷图像,缺陷细节最清晰,且双阈值分割算法能准确提取曲面玻璃中划痕缺陷。
Aiming at the difficulty of imaging and extracting scratch defects on curved glass,a method of scratch defects detection based on machine vision is proposed.Firstly,the corresponding relations among curvature,camera position and lighting Angle are deduced through mathematical modeling analysis.Then,under the guidance of this relation,defect images with different lighting angles are obtained,and a defect extraction algorithm based on double threshold segmentation is proposed.Finally,the experimental results show that the nice-detail of the scratch-defect image obtained from the deduced lighting angle is the clearest,and the double-threshold segmentation algorithm can accurately extract the scratch-defect in curved glass.
作者
王昌书
黄沿江
张宪民
卢盛林
WANG Chang-shu;HUANG Yan-jiang;ZHANG Xian-min;LU Sheng-lin(School of Mechanical and Automotive Engineering,South China University of Technology,Guangzhou 510641 China;OPT Machine Vision,Dongguan 523860 China)
出处
《自动化技术与应用》
2020年第1期134-139,共6页
Techniques of Automation and Applications
关键词
曲面玻璃
图像获取
缺陷检测
数学建模
curved glass
image acquisition
defect detection
mathematical modeling