The Examine Method Of Foamed Slag Height Based On Computer Vision

During the steel metallurgy, the phenomena of slag bubbling up is very prevalent. In the melt deoxidize craftwork of iron-bath type, the superfluous foam can send the slag overflow the stave and destroy stably operate. For obtaining the availability and stably operate of steel-making craftwork, it must to be found the slag-Fe interface. Now it is mostly grasped by worker experience, and has grate random and blindness. If we can realize the on-line measurement of foam height on site, there are huge convenience and considerable economy benefit on fieldwork.Based on a lot of experiment, using technology of image processing and pattern recognition, the Interface extraction theory and methods of automatic foamed dreg are investigated and the automatic fingerprint preprocessing identification system is developed. Aim at the study now, the key problems are discussed and the text advance an examine method of foamed slag interface based on image disposing technique. The experimental result indicates that this measure method is feasible on principle, and has directive meaning for realization of on-line measurement of foam on site. The main contents are foam image pretreatment, foam interface extraction, thin, link etc. The main jobs are follows:(1) An improved arithmetic of edge measure based on grads extremum is presented and validate. The threshold can be selected naturally according to the image in this method and difficulties in selecting the threshold are avoided. The foam image with any difference in ridge and vale gray levels can all be picked-up the edge(2) An thinning arithmetic of foam image edge is presented and validate. The experimental result indicate that the contour line of thinned foam image can be the same as the primal foam image and wipe off excrescent element information, and assure only single element width. It saves disposal speed of latter arithmetic.(3) The text using the method of least tow multiplication to implement the edge dot curvelinking of foam image. The experimental result indicate that the method not only link the primal edge outline effectually and not lost primal image information, but also simple and quick..