Online auto-detection method and system of presswork quality
Online auto-detection method and system of presswork quality
复制标题
DOI：
10.1007/s00170-006-0505-7
复制
发表时间：
2007-05
期刊：
影响因子：
--
通讯作者：
Hui-chao Shang;Youping Chen;Wen-yong Yu;Zude Zhou
中科院分区：
文献类型：
--
作者：
Hui-chao Shang;Youping Chen;Wen-yong Yu;Zude Zhou
本文研究了一种印刷品质量在线自动检测方法及系统。提出了一种多色打印实时采集图像的评价标准。综合设计和优化使得影响检测算法实时性能的关键因素变得透明。提出了一种基于子空间的配准算法，加快了搜索速度，动态刷新参考图像避免了累积偏差，提高了配准精度。为了实时获取各种缺陷的位置和几何特征，提出了一种基于动态阵列逐行扫描聚类阈值的缺陷提取与融合方法。最后，利用几何特征和统计参数对块状缺陷、离散点状缺陷和线状缺陷进行分类识别。设计了运动图像采集系统和分布式计算机处理系统。实验研究和实际运行表明，该系统能够实时检测、显示、存储和报警条纹、溅墨、吹扫、异物等缺陷。检测速度达到25万米/分钟，运行方向精度达到0.375毫米，交叉方向精度达到0.15毫米。
An online auto-detection method and system of presswork quality is studied in this paper. An evaluation criterion for real-time-gathered images of multicolor printing is proposed. Integrated designing and optimizing make it transparent the key factors influencing the real-time performance of a detection algorithm. A subspace-based registration algorithm that accelerated the search speed is presented, and the dynamic refreshing of a reference image avoided the cumulative departure and improved the registration precision. To obtain the position and geometric characteristics of various defects in real time, a method of thresholding by cluster and line-by-line scanning of a dynamic array is proposed to extract and merge defects. Finally, geometric characteristics and statistics parameters are adopted to classify and identify the massive defects, discrete dots defects, and linear defects. A motion image acquisition system and distributed computer processing system are designed. Experimental study and practical running indicates that defects including stria, splash ink, blowby, foreign bodies, etc. could be detected, displayed, stored, and warned in real time. The detection speed reached 250 m/min and the precision reached 0.375 mm in the running direction and 0.15 mm in the crossing direction.