Level set based slag detection method for BOF vessels tapping images

JIANG Jun,LI Pei-yu,TANG Ren-zhong
DOI: https://doi.org/10.3785/j.issn.1008-973X.2011.02.005
2011-01-01
Abstract:Infrared images of basic oxygen furnace(BOF) vessels tapping stream always had harsh noise,small grayscale difference between target objects and blur edges.It's difficult to achieve the target region segmentation and slag detection.In order to solve these problems,a level set based slag detection method for BOF vessels tapping images was put forward.First,according to distribution characteristics of molten steel and slag contours,the active contour model of Chan-vese were improved to construct a double-contours of curve evolution function.Then using the level set method to solve model,the segmentation of the region and its gray value could be obtained easily by minimizing double-contours of curve evolution.Finally,in terms of regions' gray values,molten steel and slag could be identified,and slag ratio be calculated.Experiment shows the high adaptability,noise immunity and accuracy of our method,which can be impressive usefully implemented in Infrared thermal imaging slag detection for BOF vessels tapping.
What problem does this paper attempt to address?