摘要
针对标签模切的凸起缺陷,采用模板对比检测方法中存在的误报和漏报的问题,文章提出了利用图像形态学处理的方法和通过提取轮廓的方法对现有的视觉检测算子改进以提升检测效果。使用HALCON软件实现模板比对、形态学和提取亚像素轮廓的三种检测的方法,并进行误报、漏报和算子处理速度进行比较。通过样本标签测试结果说明,针对样本中的凸起缺陷,改进后的方法误报和漏报为0,正确率较高。此方法对于规则物体边缘凸起缺陷的检测有较高的参考价值。
For the bump defects of the label＇s die cutting,the method using templates comparison often leads to false positives and false negatives, and the method of using image morphological processing and the method of extracting contour are put forward in this paper to Improve the existing visual inspection effect.And use HALCON software to accomplish these three visual inspection methods of template matching, morphological processing and extraction of sub pixel contour,and compare the false negatives, false positives and the processing speed of three methods. Through the sample＇s test results for convex defects in the labels, false positives rate and false negatives rate has improved to be 0. The correct rate is higher.This method of morphological processing has a high reference value for the detection of defects in the edge of the regular object.
出处
《信息通信》
2018年第1期116-119,共4页
Information & Communications
关键词
标签检测
图像处理
机器视觉
缺陷检测
Keywords： Label E〉 etection
linage Processing
Machine Vision
Defect Detection
