基于平均模板法的锆管坡口异物视觉检测研究  

Study on Visual Detection of Foreign Matter in Groove of Zirconium Tube Based on Average Template Method

作　　者：刘坚[1] 董力成 索鑫宇 LIU Jian;DONG Licheng;SUO Xinyu(State Key Laboratory of Advanced Design and Manufacturing for Vehicle Body,Hunan University,Changsha,410082,China)

机构地区：[1]湖南大学汽车车身先进设计制造国家重点实验室,湖南长沙410082

出　　处：《湖南大学学报（自然科学版）》2020年第12期53-60,共8页Journal of Hunan University:Natural Sciences

基　　金：国家科技部中日科技合作专项(2017YFE0128400);国家科技部创新方法工作专项(2016IM030300);长沙市科技重大专项(kq1804005)。

摘　　要：针对人工检测锆管内外壁坡口异物存在的效率低、准确度差等问题,设计了一种基于平均模板法的锆管坡口异物视觉检测方法.该方法采用复合光源对锆管内外壁坡口圆锥面进行成像,并对所获图像进行插值展开,基于阈值设定进行列像素灰度值的相似性比较与替换,进而设计出坡口无异物状态的平均模板,通过模板与实拍检测图像的差分定位疑似异物,最后根据实际判据规则进行异物识别.实验测试及企业应用表明,该方法能有效检测出锆管坡口处的小尺寸异物,准确度可达99.6%,检测效率为1.84 s/根,相应装备在企业运行良好.In order to solve the problems of poor accuracy and low efficiency in manual detection of foreign matters in the internal and external wall groove of zirconium tube,a visual detection method based on the average template is proposed in the paper.In this method,compound light source is used to image the conical surface of the groove on the inner and outer walls of zirconium tube,and the obtained image is unfolded into rectangle image.Based on the threshold setting,the similarity comparison and replacement of the gray value of the column pixels are carried out,and then the average template of the groove without foreign matters is designed.The suspected foreign matters are located by the difference detection between the template and the real image.Finally,the foreign matters are identified according to the actual criterion rules.Experimental test and enterprise application show that the method can effectively detect the small size foreign matters in the groove of zirconium tube,with an accuracy of 99.6%,and an detection efficiency of 1.84 s/piece,and the corresponding equipment runs well in the enterprise.

关 键 词：机器视觉 锆管 缺陷 模版匹配 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]