In sinter production, the particle size of sinter mixture has an influence on water content, permeability, sinter quality and other factors. In view of the lack of timeliness of the traditional manual method for statistical particle size, which leads to the inability to timely feedback the statistical results, based on the background of intelligence and automation, an intelligent detection method that can replace the traditional manual particle size of sinter mixture is proposed. The Canny algorithm is focused on, a new algorithm flow is built, the threshold processing is integrated with the image integration, and the distance transformation function is used to improve the morphological processing effect. At the same time, by comparing with the results of manual calibration, it is found that the improved algorithm is used to detect the image of the sintering compound, and the accuracy of the average particle size measurement is up to 90% or more, which can accurately detect the particle size of the sintering compound.