全局缺陷和局部缺陷的FPC检测算法研究

Research of detection algorithm of flexible printed circuit boards based on global and local defects

为满足柔性电路板缺陷检测在生产线上的检测要求,讨论了基于机器视觉的柔性电路板缺陷实时检测系统试验所需要的关键技术.并且针对传统的基于模板匹配的算法的速度慢、准确率低等问题,提出将柔性电路板的缺陷分为全局缺陷和局部缺陷,对全局缺陷采用直方图匹配和八连通域面积的计算,淘汰两者均存在缺陷的产品;对未被排除的产品图像采用投影匹配和相关系数的计算,实现局部缺陷的识别.改进的算法经过实验验证表明其准确率和检测速度优于传统的检测算法,符合柔性电路板生产线的检测要求.

A new method based on image analysis was discussed in order to satisfy the detection of FPC in production lines,which was the key technology needed in real-time detection systems.In view of the low speed and accuracy of the traditional matching algorithm,a classification was proposed that the defect was divided into global and local defects.Histogram matching and the area calculation of the eight-connected domain was used to judge the global defect.Then,projection matching and the calculation of correlation coefficient were used to implement the local defect detection.This new algorithm was proved in many experiments.The result proves that its accuracy and detection speed are better than the traditional algorithm.It satisfies FPC production lines,detecting demand.