Abstract:The automatic detection of the particle size distribution of the mixed ore directly guides the production and construction process, and the particle size analysis of the mixed ore based on image analysis can be used for on-site real-time detection. The boundary of massive mixed ore is accurately segmented from the image, and is disturbed by problems such as noise and adhesion of ore targets. This paper proposes a segmentation optimization method based on density transform. In the threshold image, the noise pixel and ore target adhesion problem are removed by the density statistics of the ore image neighborhood. The particle size is detected according to geometric morphological features, the particle size distribution is calculated, and the gradation curve is drawn. The experimental results show that the segmentation result and particle size distribution of the proposed optimization method are consistent with the gradation curve of the screening measurement, and can be used for grading detection at the construction site.