A profile feature mining method for Burden Surface in Blast Furnace based on monocular video images

The features for profiles of the whole burden surface topography are significant for blast furnace, which can help improve the accuracy of the charging operation. However, it remains a challenging task currently. On one hand, high-quality sensing data is difficult to obtain in the extremely harsh environment. On the other hand, postprocessing algorithm based on the low-quality data faces a series a challenges. Therefore, a novel method is proposed to obtain effective profile features based on monocular video images. First, a 3D model is constructed based on the preprocessed video image; Then, depression are extracted and main profiles are picked out; Finally, all the profiles are analyzed for the charging operations. The capability and performance of the proposed profile feature mining method are validated on practical data, which exhibits very good results.