冷延鋼板の表面欠陥検査システム

抄録

In steel industry, customers' request for better quality of product surface has recently become very keen, which inevitably requires more strict inspections of the products before shipping.
Though human eyes, which have been widely used for in-process product visual inspections, are an excellent sensor, organoleptic visual tests have drawbacks. A human inspector may overlook flaws in case of even momentary distraction. In addition, today's demands for better performance and higher throughput on production and inspection far exceed the ability of human visual tests, in most cases.
Consequently, visual tests are being replaced by ASIs (Automatic Surface Inspectors) using optical apparatuses. There still remain, however, some flaws ASIs are not able to detect because of their low signal level. We have, therefore, kept trying to detect those flaws by developing and improving signal and image processing techniques for ASIs.
In this paper, described are some of the new technologies of this kind, which have been developed or tested in Kawasaki Steel Corporation; a floating-threshold-level signal processing method, a high precise definition TV monitor system, an on-line ASI performance test system, and Fractal analysis for the scanning waveform of ASIs.