The image recognition of mobile robot is to extract the effective target information. The essence of image extraction is image segmentation. By extracting and distinguishing planar objects and three-dimensional objects, we propose two new algorithms. The color image is extracted by using CIE Lab Space. Then we propose a comparison method through the collection of two image samples. According to the principle of geometric distortion in the geometric space, we can easily distinguish the planar object in the environment. Therefore, Experimental results show that the combination of these two methods is accurate and fast in the color image recognition.