篇名 | Multivariate Positioning and Dimension Measurement Technology Based on Template Matching |
---|---|
卷期 | 31:6 |
作者 | Xin-Yu Hu 、 Xue-Sheng Li 、 Tong Ye 、 Xian-Guo Liu 、 Dao-De Zhang |
頁次 | 091-103 |
關鍵字 | edge detection 、 IC chip inspection 、 machine vision 、 template matching 、 visual detection system 、 EI 、 MEDLINE 、 Scopus |
出刊日期 | 202012 |
DOI | 10.3966/199115992020123106008 |
In the process of semiconductor chip packaging, to solve the problem of multiple positioning and precise size measurement, this paper proposes a visual positioning and measurement technology based on template matching. Firstly, the identification points are obtained from the calibration plate and the internal and external parameters of the camera calibration are calculated. Then, based on the comparative analysis of different template-matching and positioning technologies, the shape-based template-matching method is adopted to quickly match the shape model of the best target chip, and the pyramid layered search strategy is applied to realize the online real-time positioning and identification of the chip under different lighting conditions. Finally, the gauss fitting sub-pixel edge detection method was designed to obtain the sub-pixel edge contour of the target chip, and the precise measurement of pin spacing and external dimension of the chip was realized. Experiments show that the detection system can realize accurate identification of different kinds of target chips under complex conditions such as occlusion and stack, and has low requirements for lighting environment. It can satisfy real-time and accurate positioning and measurement of multiple chips, and the actual measurement error of the system can reach 0.16%.