Research On Iron-Ore Pellets Granularity Detection Method Based On Machine Vision

Iron-ore pellet plays an important role in iron and steel smelting with good strength,high iron content and good reducibility.Pellet size is an extremely important quality index for iron-ore pellets.However,the granularity of pellets is mainly measured by traditional sieving method and manual detection method.These two methods not only have low detection efficiency and poor accuracy,but also the grain size information can't online feedback in real-time,seriously affecting the quality of pellet.In this dissertation,the machine vision detection technology is applied to the iron-ore pellet size detection,which is the main research direction of the iron-ore particle size detection.The pellet production environment is complex,making the original pellet image contains so many noises so that it must be pretreated.This dissertation uses Butterworth filter method,which could eliminate the noise and protect the edge information of the image;The images are reconstructed by opening and closing so that the light and dark details in the iron-ore pellet image are eliminated,and no new boundary or boundary displacement phenomenon is created after reconstruction,and the maximum and minimum values in the target area are amended.Iron-ore pellet image is a typical circular stacked particle image,to achieve effective segmentation is the key to ensure the accuracy of size measurement.This dissertation analyzes the difficulty of image segmentation of pellets,found that the traditional methods can not meet the requirements of pellets image segmentation,therefore this dissertation proposes a watershed segmentation algorithm based on adaptive tag.The simulation results show that this method can separate the target pellet accurately,which can solve the problem of circular stack of particle images such as occlusion and adhesion in the image,and overcome the over-segmentation of the traditional watershed algorithm.After the image segmentation is completed,the solid region of the binary image is marked by the line mark method,and the target region that can not correctly reflect the particle granularity information is deleted by the proposed index of FUR,which is used to measure roundness,to ensure the accuracy of the detection results.Then,the coefficients of the image detection system are calibrated and the actual area of the pellets target in the iron-ore image is extracted,and then obtain the particle size value of the pellet.Finally,the results proved that the proposed method of particle size detection based on machine vision has good accuracy and reliability.Compared with the traditional screening method,the proposed method is more accurately and conveniently and can realize on-line detection,so that it will play an important role in improving pellet production quality.