自动光学检测的彩膜分区检查与判定  

Color filter partition inspection and judgment of automatic optical inspection

作　　者：张铁轶 余道平 王野 刘超强 张祥 

机构地区：[1]北京京东方显示技术有限公司,北京100176

出　　处：《液晶与显示》2014年第1期34-39,共6页Chinese Journal of Liquid Crystals and Displays

基　　金：安徽省2012科技计划资助项目(No.1206c0805007)

摘　　要：为了提高自动光学检测系统检测彩膜质量的准确率,降低误判,通过分析由红绿蓝三色组成矩阵结构和Line CCD获取的彩膜图像特征,在五点周期比较检测的基础上设计一种彩膜缺陷的检测方法。该方法通过对彩膜的机械与光学对位后,通过五点比对进行缺陷的检测。Line CCD从红绿蓝3个不同的区域获取不同的灰阶值,分别设置3个区域的灰阶值范围和该区域的判定阈值;最后根据缺陷与正常点灰阶差值的不同对缺陷进行分类,赋予相应的缺陷代码。基于以上3个步骤,实现对整张彩膜的缺陷检测。实验结果表明,采用五点比较、分区检测和缺陷分类相结合的方法,缺陷检出的准确率可以提高至99.6%以上。In order to improve the inspection accuracy and reduce misjudgment rate for the color filter quality inspection in an automatic optical inspection system,a defect inspection method for color filter was proposed based on analyzing the color filter image acquired by the red, green and blue color matrix structure and Line CCD. Three steps were used to detect the defect of color filter. Defects of color filter were inspected firstly by the five points comparison method after mechanical and optical alignment for color filter.Then, according to Line CCD obtained by three different grayscale values from the red, green and blue area, the determination threshold are respectively provided in the three regions. Finally, the defects were classified depending on the grayscale difference in the defect and the normal point, and the corresponding defect code is given. The experiment results show that the proposed method composed of the five points comparison method, partition detection, defect classification can make inspection accuracy reach more than 99.6%.

关 键 词：自动光学检查 彩膜 灰阶 缺陷 

分 类 号：TH865[机械工程—仪器科学与技术] TH691.9[机械工程—精密仪器及机械]