基于机器视觉的半导体表面缺陷检测研究

半导体的广泛应用使得人们对其质量要求越来越高。半导体生产过程中塑封表面缺陷的检测得到更多生产厂家的重视。本文对SOT-23封装式半导体的表面塑封缺陷进行了研究,改进并优化半导体质量缺陷检测系统,通过边缘检测算法提取半导体管体的塑封边缘,确定被检管体的位姿,缩小检测范围,使用基于边缘点的模板匹配算法来判断字符的完整性,最后利用差影法准确地检测出塑封缺陷。经工厂试用后,结果显示此系统能有效地检测出表面质量缺陷。目前此系统已在工厂正式使用。

Study on Semiconductor Surface Defect Detection Based on Machine Vision

More attentions have been paid on semiconductor quality with the wide application of semiconductors. Production manufacturers want a faster method to detect the package surface defects of semiconductors. This paper developed a system to detect the surface defects for SOT-23 package semiconductors. The semiconductor plastic edge was extracted by edge detection algorithm to conform the position of the body. The detection area would be reduced and the template matching algorithm base on edge points was used to estimate the integrity of the character on semiconductor package surface. Finally, the package defects can be detected exactly by different algorithms. The system has been proved effective in defect detecting and is employed in factories.