Abstract: Image segmentation is the foundation of computer image analysis and understanding.And color is an important way for image segmentation.In this paper, a color image segmentation algorithm based on color learning is introduced. The algorithm is performed by calculating the color features of the normalized tristimulus image r, g, and b, and also by template matching. By using the algorithm, the negative effects of uneven lighting and shading of the image on segmentation can be overcome, and several areas can be segmented simultaneously according to the colors being assigned. The algorithm can be implemented simply and fastly, and it is an effective color image segmentation method.