Novel Recognition Method of Blast Furnace Dust Composition by Multifeature Analysis Based on Comprehensive Image-Processing Techniques

The traditional artificial recognition methods for the blast furnace dust composition have several disadvantages, including a great deal of information to dispose, complex operation, and low working efficiency. In this article, a multifeature analysis method based on comprehensive image-processing techniques was proposed to automatically recognize the blast furnace dust composition. First, the artificial recognition and feature analysis, which included image preprocessing, Harris corner feature, Canny edge feature, and Ruffle feature analysis, was designed to build the template image, so that any unknown dust digital image could be tested. Second, the composition of coke, microvariation pulverized coal, vitric, ash, and iron from dust would be distinguished according to their different range of values based on the multifeature analysis. The method is valid for recognizing the blast furnace dust composition automatically, and it is fast and has a high recognition accuracy.