Lens Quality Inspection using Image Processing and Machine Learning

Abstract

This research proposes a system to inspect defective lenses with a polarization technique by using image processing and machine learning. Currently, a skilled operator checks the lens quality with the polarization method by eye and decides whether or not a lens is good (OK) or not good (NG). A 'not good' lens has a circle or a line appearing in the stress pattern of the lens. This research designs and develops a lens quality checking system with machine learning by simulating and prototyping a machine to experiment and collect persistent data, using the camera to capture and analyze images with image processing and machine learning techniques to decide on the lens quality in the computer. The experimental results show that the proposed system with a trained model with data augmentation and image preprocessing can achieve performance testing with 97.75% accuracy.