张德海,祝志逢,李艳芹,黄子帆,马选雄,许宸语,刘祥.基于机器视觉的二维图像质量缺陷检测研究进展[J].包装工程,2023,44(23):198-207.
ZHANG De-hai,ZHU Zhi-feng,LI Yan-qin,HUANG Zi-fan,MA Xuan-xiong,XU Chen-yu,LIU Xiang.Research Progress of Two-dimensional Image Quality Defect Detection Based on Machine Vision[J].Packaging Engineering,2023,44(23):198-207. |
| 基于机器视觉的二维图像质量缺陷检测研究进展 |
| Research Progress of Two-dimensional Image Quality Defect Detection Based on Machine Vision |
| 投稿时间：2023-09-12 |
| DOI：10.19554/j.cnki.1001-3563.2023.23.024 |
| 中文关键词: 机器视觉 印刷质量 缺陷检测 图像处理 |
| 英文关键词: machine vision printing quality defect detection image processing |
| 基金项目:国家自然科学基金青年项目(52006201);国家自然科学基金面上项目(52275295);郑州轻工业大学横向项目(JDG20210045) |
|
| 摘要点击次数: |
| 全文下载次数: |
| 中文摘要: |
| 目的 机器视觉图像处理技术是近年在图像处理领域发展起来的一门新兴边缘交叉学科，二维图像的质量检测是印刷行业中必不可少的环节，分析基于机器视觉的二维图像质量缺陷检测流程，探索影响基于机器视觉的二维图像质量缺陷检测精度的相关因素，为后续研究印刷品的二维图像自动化检测和质量控制提供参考。方法 在此基础上，围绕图像预处理中的灰度转换、噪声过滤、固定阈值分割、自适应阈值分割、Otsu法及边缘检测，对图像配准中的基于灰度统计信息分布配准方法、基于特征的图像配准方法进行总结，然后归纳分析图像的缺陷提取和分类。结论 以实际例子对上述研究内容进行了提炼，通过图像预处理中的噪声过滤为后续缺陷提取提供清晰图像，减少伪影干扰;通过图像预处理中的灰度变换、阈值分割、感兴趣区域提取减少系统处理时间，为实现高效的缺陷检测奠定了坚实的基础;通过图像配准消除了机械振动引起的图像位置偏移，确保后续缺陷提取的准确性;通过图像缺陷提取和分类帮助印刷企业找出生产问题，提供有针对性的改进措施，可为生产高质量产品提供支持。 |
| 英文摘要: |
| Machine vision image processing technology is an emerging fringe cross-disciplinary discipline developed in the field of image processing in recent years, and the quality inspection of two-dimensional images is an essential link in the printing industry, analyzing the quality defect detection process of two-dimensional images based on machine vision, exploring the relevant factors affecting the accuracy of two-dimensional image quality defect detection based on machine vision, and providing references for the subsequent research and development of automated inspection and quality control of two-dimensional images of printed materials. On this basis, around the gray scale conversion, noise filtering, fixed threshold segmentation, adaptive threshold segmentation, Otsu method and edge detection in image preprocessing, the gray scale statistical information distribution based alignment method and feature based image alignment method in image alignment were summarized, and then the defect extraction and classification of images were summarized and analyzed. The above research content is refined with practical examples. Noise filtering in image preprocessing is used to provide clear images for subsequent defect extraction and reduce artifact interference. Gray scale transformation, threshold segmentation and region of interest extraction in image preprocessing are used to reduce system processing time, laying a solid foundation for efficient defect detection. The image location offset caused by mechanical vibration is eliminated by image registration to ensure the accuracy of subsequent defect extraction. Image defect extraction and classification can help printing companies find production problems and provide targeted improvement measures for the production of high-quality products, thus providing important support. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
渝公网安备 50010702501716号