Research On The Pellet Particle Size Detection System

In the process of blast furnace smelting, the high quality pellets particles can notonly improve the permeability of charging concentrate, reduce production energyconsumption, but also reduce slag quantity at the same time. And an importantindicator in determining the quality of pellets particles is it’s size. The traditionalscreening method to detect pellet size has many disadvantages such as, low detectionefficiency, poor accuracy and so on. In order to overcome these shortcomings, thispaper applies image processing technique to the pellet size detection, which will notonly avoids the influence of the testing personnel subjective factors, but also improvesthe testing efficiency and accuracy.To solve the problems of much noise in the pellet particle image and the pelletparticle edge is not clear, This paper adopts the bilateral filtering method, which cannot only eliminate the noise in the image effectively and can well retain the edge of thepellets at the same time; And the contrast limited adaptive histogram equalizationmethod is used to increase the contrast between the pellets particles and thebackground, highlight the edge information of pellets particles;Aiming at pellet particle adhesion in the image, the watershed algorithm based onadaptive marker is used to segment the image, First we use morphological opening andclosing reconstruction operation creates and extract the local maximum value of thePellets particles and make foreground markers, then isolates the adhesion pellets bywatershed transform, avoids the over-segmentation phenomena and obtains bettersegmentation effect; For the problem of the pellet size of the incomplete pellet mayprobably partial loss, we use Hough transform method to edge fitting the incompletepellets and detect the size of the incomplete pellets accurately.Compared with the actual pellet size, image processing method to detect pelletsize is faster and has higher accuracy and can reflect the distribution of grain size ofthe pellets intuitively and with detail.