液晶模组自动光学检测算法研究  

Research on automatic optical detection algorithm of liquid crystal modulee

作　　者：李海霞[1] 张衡 Li Haixia;Zhang Heng(Lanzhou Petrochemical College of Vocational Technology,Lanzhou 730060,China;Aerospace Science and Technology Intelligent Robot Co.,L.td,Beiing 100070,China)

机构地区：[1]兰州石化职业技术学院,兰州730060 [2]航天科工智能机器人有限责任公司,北京100070

出　　处：《电子测量技术》2020年第13期95-98,共4页Electronic Measurement Technology

基　　金：甘肃省高等学校创新能力提升项目(2019A-199)资助。

摘　　要：主要针对3~8 in的手机液晶模组,实现液晶模组表面各种类型缺陷的全自动化检测。缺陷检测过程基于机器视觉,采用图像处理技术作为核心内容,并结合了视觉定位系统。视觉检测软件首先提取被测图像ROI区域,然后做实值Gabor滤波、双阈值二值化和图形形态学等预处理;接着,采用二维直方图斜分法定位图形的边缘区域,并对边缘区域进行模糊增强处理;采用顺序形态学检测图像边缘,利用结构元素和百分位参数得到子图像,对边缘子图像进行加权融合处理,得到最终边缘,以此来计算出面积、形状等缺陷。缺陷检测实验结果表明,缺陷陷识别准确率大于95%,漏检率小于0.2%,检测时间小于25 s,达到设计的目标要求。It is mainly used for 3~8 in mobile phone LCD module to realize automatic detection of various types of defects on the surface of LCD module. The defect detection process is based on machine vision with image processing technology as the core content, combined with the visual positioning system. The visual detection software first extracts ROI of the tested image, and then does real value Gabor filter, bi-threshold binarization and image morphology preprocessing;then, locates the edge region of the image by two-dimensional histogram skew method, and enhances the fuzzy edge region;detects the edge of the image by order morphology, obtains the sub-image with the help of structure elements and percentile parameters, and merges the edge sub-image with the linear weighting to get the final edge, so as to calculate the area, shape and other defects. The defect detection experiment results show that the accuracy rate of defect identification is greater than 95%, the missed detection rate is less than 0.2%, and the detection time is less than 25 s, which meets the design target requirements.

关 键 词：自动光学检测 表面缺陷 机器视觉 液晶模组 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]