The invention discloses a mineral rock granularity detection method based on image processing. The method comprises the steps of 1, establishing of a mineral rock granularity database; 2, image acquisition; 3, image synthesis; 4, luminance conversion; 5, gray scale transformation; 6, determining of the Gaussian standard deviation of the improved Gaussian filter. 7, removing noise; 8, morphologicalreconstruction; 9, distance transformation; 10, forcing minimum; 11, image segmentation; 12, image calibration; and 13, information output. The invention has high detection efficiency, and can filterthe image according to individual properties of different images. After the improved Gaussian filter processing and noise removal, the image of the ore rock is clearer than the original image, the noise generated in the image of the ore rock is eliminated, the obvious gray step between the ore rock particles and the background is eliminated, and the boundary information of the ore rock is reserved. So the accuracy of particle size detection is improved.