부산대학교 도서관

This paper presents an implementation of an AI-based vision inspection system for semiconductor process quality control. The hardware of the optimal vision inspection system consists of a conveyor belt structure transport system and a real-time high-speed line scan camera device module. And the software system consists of a trigger- based image acquisition module, an image processing module, and an AI module. The AI module for semiconductor process quality control, which is the core function of the proposed system, uses an image recognition technology based on supervised learning, so it is carried out to recognize and classify images by extracting features from labeled data and learning to classify images. The implemented system is useful to the improve semiconductor quality control.