Machine-vision based detection for sheet metal industries

In this work, the authors present a machine-vision system (MVS) whose features are able to detect small protrusions (diameter around hundred /spl mu/m) or peaks over external car parts, for the performance monitoring of steel sheet machining processes in the automotive industry. Trials made in this research prove the feasibility of implemented MVS based on two-dimensional imaging through a framegrabber. Also, algorithms for peaks detection are performed in a Pentium II PC, and developed by an object oriented programming tool called IMPULS VISION XXL SDK. The paper introduces the main fault for inspection over steel sheets devoted to external car parts. Relevant features for MVS are extracted from trials made in steel sheet machining shopfloor, called MVS and scanning ones. Algorithms for peaks detection are also presented. Results and new guidelines of this work are explained, as are conclusions and future works.