On-line machine vision system for detect split defects in sheet-metal forming processes

Abstract

In this paper, we present an automatic system designed for detect the presence of split defects in sheet-metal forming processes. The image acquisition system includes basically a CCD progressive camera and a diffuse illumination system mounted on the end-effector of a 6-dof robot. The inspection-robot displaces the image acquisition system over the pieces proceeding from the sheet-metal forming line. The recognition, positioning and the later inspection are realized as the pieces are moving on a conveyor belt. To realize the inspection, the acquired images are restored using a Markov random field model. Defect detection is carried out using a valley detection algorithm. To realize the recognition and to determine the precise position, we have used an appearance-based method, based on a principal component analysis (PCA)