汽车车身涂膜缺陷的计算机视觉检测方法.pdfVIP

第 42 卷 第 12 期 天 津 大 学 学 报 Vol.42 No.12 2009 年 12 月 Journal of Tianjin University Dec. 2009 ▋ 汽车车身涂膜缺陷的计算机视觉检测方法 陈东祥，刘 磊，韩鸿志 (天津大学机械工程学院天津市先进制造技术与装备重点实验室，天津 300072) 摘 要 ：将计算机视觉检测技术应用于汽车车身表面涂膜缺陷的检测，根据汽车涂膜及涂膜缺陷的特征，提出了侧向 照明方式，获取了特征明显的涂膜缺陷图像．针对汽车涂膜缺陷边缘模糊，传统的微分检测算子无法有效检测的情况， 提出了基于图像中心化和图像去背景操作相结合的检测分割算法，算法原理简单，性能稳定，能够有效检测出绝大多数 涂膜缺陷的模糊边缘． 关键词 ：涂膜缺陷；计算机视觉；模糊边缘；模式识别 中图分类号：TP391 文献标志码 ：A 文章编号 ：0493-2137 (2009)12-1130-05 Detection Method of Car Body Painting Defects Based on Computer Vision Technology CHEN Dong-xiang ，LIU Lei ，HAN Hong-zhi (Tianjin Key Laboratory of Advanced Manufacturing Technology and Equipment ，School of Mechanical Engineering ， Tianjin University ，Tianjin 300072 ，China) Abstract ：In this paper ，computer vision technology was applied to detect the car body painting defects. Based on the characteristics of car painting coat and its defects ，side direction illumination mode was developed in order to obtain the distinguishable painting defect images. Meanwhile considering that conventional detection operators are not effective to detect fuzzy edges of car painting defections ，a defect segmentation algorithm was put forward based on the image centerization and morphological operation. The algorithm performs stably with simple principle ，and can effectively detect most of painting defection fuzzy edges. Keywords ：painting defect ；computer