摘要
为了实现PCB的孔位信息在线检测,依据机器学习理论中机械学习法模型,抽象出整个图像检测过程中的要素,建立了对应的自动光学检测的学习模型.针对工件定位时初始位置的微小错位可能导致的自动检测意外中止,开发了三类专门的图像检测算法,减少了自动检测出错几率.在BolandC++环境下,开发设计了相关学习类,并按照设计的检测学习过程流程和检测执行过程流程,开发了对应的检测软件,实现了PCB孔位信息的流水线自动化全检.该研究结果大幅度提高了PCB板检测效率,有助于改善其制造品质,对于PCB其他瑕疵检测具有参考意义.
In order to achieve on-line inspection on PCB holes,this study extracts the parametric elements during a complete process of image inspection,and establishes an auto optical measuring learning model for the PCB holes auto-inspection based on machine learning model.In practices,using traditional measuring method,a minor initial position error may cause unexpected inspection termination.Therefore,three image measuring algorithms have been developed to overcome this effect.In addition,under Boland C++ environment,the TAutomethodinspection and TAutoinspection classes have been designed using learning model.A measuring software has been developed according to machine learning and auto-inspection execution process.This measuring system is used for the PCB on-line measurement to increase measuring efficiency.This study significantly improves the PCB measuring efficiency,and improves manufacturing quality.It also provides a useful reference for other PCB defect detections.
出处
《辽宁工程技术大学学报(自然科学版)》
CAS
北大核心
2012年第1期93-97,共5页
Journal of Liaoning Technical University (Natural Science)
基金
浙江省科技厅自然科学基金资助项目(2008C0025)
安徽省教育厅自然科学研究项目(KJ2011B036)
关键词
机器学习
人工智能
在线检测
自动光学检测
线侦测法
同心圆区间侦测法
区域分割法
视觉检测
machine learning
artificial intelligence
on-line measurement
AOI
line detection method
concentric circle area range detection method
area segmentation method
visual inspection