Abstract: Considering the low segmentation speed of color image segmentation based on clustering method, a color image segmentation algorithm was proposed based on spherical granular computing. For a color image, each pixel point was represented as a sphere with the center RGB and radii 0, the union operator between two spherical granules was designed, the granularity threshold was used to unite two sphere granules and obtain the sphere granules with different granularities. The RGB of pixels belonging to the sphere granule was replaced by that of the sphere granule’s center. The experimental results showed that the spherical granular computing segmentation algorithm was stable and speeded up 6 times and 34 times compared with Kmeans and FCM segmentation algorithms,respectively.