バレル研磨面の知的認識法

Abstract

This paper describes an intelligent visual system for evaluating barrel-finished surfaces using a fuzzy-neural network. If the features of image processing are recognized by the vision system and fuzzy-neural network, the degree of surface roughness, brilliance, and/or types of barrel-finishing conditions will be clarified in relation to surface image data and the surface integrity of barrel-finished surfaces is cleared by correlation analyses between features of surface image processing dataThe objectives of the paper are to analyze the image of barrel-finished surfaces using both texture and color texture analyses, and to construct a fuzzy-neural network system able to evaluate surface roughness, brilliance and types of finishing.It was found that the features of image data of barrel-finished surfaces is decided by texture analyses. Results show it is possible to construct a fuzzy-neural network system composed of surface image features as input data and the degree of surface roughness, brilliance, and/or types of finishing as output data.