In hot slab continuous casting (CC) process, the Machine Vision (MV) inspection method for slab surface defects has been becoming a research focus with the development of charge-coupled device (CCD) camera imaging technologies. To acquire a informative and reliable defects image is a prerequisite that means a maximum defects detection rate and minimum false recognition in the next image processing stage. In this paper, based on the CCD imaging principle and definition evaluation algorithm, the slab surface radiation feature at high temperature and CCD optical imaging properties had been analyzed under the focal plane, meanwhile, the CCD image definition of optical integral time controlling method had also been studied, which the defect image variance (IV) function, image average gradient (IAG) and information entropy (IIE) were calculated and analyzed. In the end, the image definition controlling algorithm for hot slab surface ROI imaging that adopted IIE evaluation algorithm, based on CCD integral time imaging controlling principle, had been proposed. This method also can support practical value for other machine vision engineering.
Read full abstract