Automation of Visual Inspection in Polishing Process Using Image Processing

Quality Judgement Using Neural Network

Abstract

This research is aiming at developing a visual inspection method of polishing surface using image processing for automation of polishing process. The method to judge polished condition by pattern recognition using neural network has been developed. Since the advance specification of feature value is not required, the intuitive judgment has become possible, which is closer to experts than the previous methods using machine learning and so on. The improvement in the accuracy of distinction between milling and polished surfaces was actually confirmed by some polishing experiments.