摘要
圆网印花机的对花系统在图像处理过程中需满足实时性要求。结合研究课题的特点给出了一种基于分块匹配策略的快速匹配算法,其目的是为了得到错花图像的位置误差矢量,为以后的闭环控制打下基础。该算法结合了点模式匹配算法及块匹配算法的思想,可分为离线的模板制作和在线的分块匹配2部分。分块匹配策略减少了数据运算量,提高了运算速度,实验表明,该算法较基于灰度值的全局匹配算法大大提高了匹配速度,是一种实用有效的快速匹配算法,不仅可以达到高的精度,而且通过分块处理,提高运算速度,可以满足印花制品在线检测的要求。
In image processing, the registering system of Rotary Screen Printing Machine （RSPM） needs to meet the demand of real-time inspection. This paper presents a fast matching algorithm based on pattern inspection system research of RSPM, its goal is to obtain the position vector of off-pattern-image, builds the foundation for the later closed-loop control. The algorithm combines the point pattern matching algorithm and the block-matching algorithm. Block matching strategy reduces the data computing and improves the operation speed, the simulation results show that it outperforms traditional matching algorithms of grey level based full search strategy, and it is a practical and suitable matching algorithm, it can not only achieve high accuracy, but also improve computing speed through the block matching strategy. The online inspection of RSPM can be realized.
出处
《纺织学报》
CAS
CSCD
北大核心
2008年第7期112-116,共5页
Journal of Textile Research
基金
国家自然科学基金资助项目(60775052)
教育部新世纪优秀人才支持计划(NCET-04-415)
教育部科技创新工程重大项目培育资金项目(706024)
上海市国际科技合作基金项目(061307041)
关键词
特征点
在线检测
快速匹配
圆网印花
图像处理
feature points
on-line inspection
fast matching
rotary screen printing
image processing
