Automatic slag characterization based on hyperspectral image processing

Nowadays, there are certain tasks that are not automated and still rely on visual inspection of an expert. In the case of the steel industry, the estimation of the deoxidation and desulfuration of steel is done at the Ladle Furnace by an expert and it is latterly analyzed in a laboratory by the use of modern spectrographs. However, the automation of this process does not constitute a trivial task. In this paper we propose a novel method for steel slag characterization, in order to automate the previous online estimation process. This paper presents the used algorithm, which is based on computer vision based hyperspectral imaging and preliminary results that forecast a promising use of optical hyperspectral technologies for this industrial application.