Abstract
Particle size distribution (PSD) of green pellets is critically important in pelletizing process because it greatly affects the quality of iron ore oxidized pellet as well as the efficiency of blast furnace process. Image-based methods are effective in automatic measurement of PSD, but are limited in their capacity in segmenting overlapped pellets and computation efficiency. To overcome these shortcomings, a novel image segmentation algorithm is proposed in the present work, with which overlapped pellets in the image can be well separated and thus the PSD of iron green pellets can be measured online with good accuracy. The proposed algorithm first identifies the markers of each pellet directly from the greyscale image by a dual morphological reconstruction method, then a circle-scan method is proposed to split the overlapped pellets and measure the diameter of each segmented pellets. The proposed methods were verified by experiments and compared with methods available in literature as well as manual sieving results. It was demonstrated that the proposed algorithm achieves better segmentation performance and has an accuracy of 94.3%. The proposed measuring system has already been tested in a local steel company. Results show that the robustness and computing efficiency of the proposed image-based method is sufficient to satisfy the requirement for online PSD measurement in harsh environment of the pelletizing process.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have