| 董正天,刘斌,胡春海,高明坤,李沛航.基于机器视觉的丝印样板表面缺陷检测方法研究[J].高技术通讯(中文),2020,30(12):1309~1316 |
| 基于机器视觉的丝印样板表面缺陷检测方法研究 |
|
| |
| DOI：10.3772/j.issn.1002-0470.2020.12.012 |
| 中文关键词: 机器视觉； 丝印样板； 缺陷检测； 金字塔算法； 归一化互相关（NCC）； 轮廓分析 |
| 英文关键词: machine vision, screen printing template, defect detection, pyramidal algorithm, normalized cross-correlation (NCC), contour analysis |
| 基金项目: |
|
| 摘要点击次数: 1944 |
| 全文下载次数: 1344 |
| 中文摘要: |
| 针对丝印样板表面缺陷检测采用人工目测法其检测效率低且漏检率高等问题，提出了一种基于机器视觉的丝印样板表面缺陷检测方法。对于边缘断裂缺陷，首先采用基于图像金字塔和归一化互相关（NCC）函数相结合的算法定位可能产生缺陷的边缘区域，然后在边缘区域生成一条灰度值扫描线，根据扫描线上的灰度值判断是否存在边缘断裂缺陷。对于圆度不完整缺陷，首先按照种子搜索、曲线追踪、曲线连接和轮廓选择的步骤提取出圆形轮廓，然后基于轮廓像素点拟合得到理想的圆形轮廓曲线，最后逐点比较提取的轮廓与拟合圆形曲线上对应点之间的距离，并根据设定的距离范围来判断被测目标轮廓是否存在圆度不完整的缺陷。实验结果表明，该方法实现了对丝印样板表面边缘断裂缺陷和圆度不完整缺陷的检测，提高了检测效率和准确率，缺陷检测的综合准确率达到94.6%。 |
| 英文摘要: |
| Aiming at the problems of surface defect detection of screen printing template using the method of manual visual inspection, i.e. low detection efficiency and high missed detection rate, a method based on machine vision for surface defect detection of screen printing template is proposed. For edge fracture defects, an algorithm based on image pyramid and normalized cross-correlation （NCC） function is first used to locate the edge region that may produce defects, and then a gray value scan line is generated in the edge region, according to the gray on the scan line. The degree value determines whether there is an edge fracture defect. For the roundness incomplete defect, the circular contour is first extracted according to the steps of seed search, curve tracking, curve connection and contour selection, and then the ideal circular contour curve is obtained based on the contour pixel point fitting. Finally the distance between the extracted contour and the corresponding point on the fitted circular curve is compared point by point, and whether the measured target contour has defects with incomplete circularity is judged according to the set distance range. The experimental results show that the method can detect the defects of surface edge defects and incompleteness of the roundness of the screen printing template, effectively improve the detection efficiency and accuracy, and the comprehensive accuracy of defect detection reaches 94.6%. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
