Feature Extraction Algorithms Based on Coal Injection Image of Video in the Blast Furnace

The status of the blast tuyere from monitoring video is usually estimated by artificial observation, but it is difficult to ensure the consistency and uniformity of the flow of pulverized coal injection. In this paper, image analysis of video recordings is presented and discussed as a novel means to estimate the instantaneous coal flow. Appropriate feature extraction algorithms based on coal injection image of video are the key technology in this method. According to the field process characteristics and two-dimensional image of coal group, feature extraction algorithm based on area statistical is proposed to estimate the flow. In addition, in order to achieve the estimation of the space volume of coal group, three kinds of volume feature extraction algorithm (Grey weighted method, the equivalent ellipse method and the axis rotation method) are explored. The experiment shows that these algorithms can reflect the flow changes accurately, and the errors of these algorithms are all within ±15%, meeting the requirements of the field well.