一种印刷机械故障标识检测新方法的研究

A new detection method for fault identity of printing machine

提出了一种结合细化算法和最小二乘法拟合的快速直线检测方法,以用于印刷画面中的标识检测,为印刷偏差自动计算提供支撑。该方法具备运行速度快、精度高的特点,符合实际印刷生产中即时性高的要求。首先,通过ZS算法实现了图像线条的细化,提高识别精度,结合最小二乘法拟合各个像素点以实现倾斜矫正;其次,利用灰度积分投影方法对直线的位置关系进行分析,实现了直线的定位;最后,通过印刷品实验验证方法的准确性。与Canny算法对比发现,本文计算的均值更接近理论值,标准差更小,提高了定位精度。相关研究为印刷故障检测标识的在线测量提供了理论基础。

A rapid line detection algorithm combining fast thinning algorithm（FTA）and least square method is put forward,which can be used to detect the lines local property in printed pictures,thereby providing strong support for automatically calculating printing registration deviation.This algorithm has various advantages,such as fast operation speed,high accuracy and practicality.The research mainly contains the following three parts.Firstly,line thinning is realized using ZS algorithm,with slant correction carried out through fitting the pixel points in the lines by the least square method.Secondly,rapid line detection is realized using gray-level integration projection.Finally,through experimental contrasts,it is found that the calculated average values are closer to theoretical ones;with standard deviation decreased,the conformity of data sets is improved and higher line location accuracy is obtained.