Design and Implementation of Optical Lens Defect Detection and Classification System

The glass lens production process requires cutting, grinding and polishing, it is easy to cause lens defects, and the production quality needs to be maintained through inspection. In the past, people relied on manual methods to detect the defects, but it causes inconsistent quality standards and costs a lot of manpower to carry out this operation. In this paper, we developed a proper image acquisition system and combine deep learning techniques and image processing algorithms to implement stringent inspection criteria based on different inspection specification requirements. Finally, a defect detection and classification system for glass lenses was successfully implemented with a defect recognition accuracy of over 95% by comparison and verification.