机器视觉在LED芯片外观检测系统中的应用

Inspection for Appearance Defects of LED Chip by Machine Vision

机器视觉系统通过对芯片外观进行图像检测处理能迅速判断出外观缺陷的位置及其类型,有效克服了传统人工目检方式造成芯片外观部分缺陷的漏检问题,提高了缺陷检测效率。本文利用基于机器视觉的芯片外观检测系统,针对红黄和蓝绿两种发光颜色不同的LED芯片产品在生产制造过程中出现的外观缺陷进行自动检测,实验表明机器视觉检测系统对于芯片的外观缺陷检测有很好的效果。通过输出数据和晶圆图的方式可以快速地判断出缺陷类型,利用加严卡控标准对TS0408产品同轴花斑缺陷检测的异常率从4.61%上升至25.69%,提升了21.08%。

The machine of vision system can quickly identify the location and type of appearance defects through image detection of chip appearance,which effectively overcomes the missing defects caused by traditional manual visual inspection,and improves the efficiency of defect detection.Here,the appearance inspection system based machine vision is used to automatically detect the appearance defects of red-yellow and blue-green LED chips in the process of manufacturing.The experiments show that the machine vision inspection system has a good effect on the appearance defects detection of chip.The defect types can be quickly determined by outputting data and wafer map.The abnormal rate of TS0408 coaxial spotted defect detection increased from 4.61%to 25.69%by using stricter standards,increasing by 21.08%.