In order to solve the problem that pellets particle size information cannot be obtained accurately and instantaneously in the production process of pellets, a particle size detection method for pellets based on image enhancement is proposed after the study of the existing detection methods for particle size of each pellet. In this method,the real-time pellets image is obtained by using an industrial camera,in view of the characteristics of low brightness and high noise of the acquired images,the improved Multi-scale Retinal Enhancement Algorithm with Color Recovery (MSRCR) combined with Log transform is used for image enhancement;and then,through guided filtering,the noise is removed while retaining the information at the edge of pellets;finally,the Hough transform of introduced threshold segmentation is used to quickly detect the number of pellets in the image,and then calculate and count the particle size of each pellet.The results show that the accuracy of the method for the detection of the number of each pellet is 92.03%,and the accuracy of the detection of the number of each qualified pellet is 91.25%,which can meet the requirements of pellets production.The results have a positive effect on reducing the labor intensity of workers,and improving the production efficiency of pellets and the pass rate.