Intelligent visual evaluation system for mirror finished surfaces

摘要

This paper describes the intelligent visual evaluation system for mirror finished surfaces of heat treated steel using a neural network. The automated evaluation system for mirror finished surfaces by an image processing has almost not been reported in the past, because of the complexity of image processing. If the features of image processing data are recognized by a vision system and a neural network, the degree of surface roughness and/or kinds of finishing methods will be clarified in relation to the image data of surfaces, and also the surface integrity of mirror finished surfaces will be clarified by the correlation analyses between the features of image processing data of surfaces.