Abstract: This paper presents a fast adaptive segmentation method for mobile robot based on histograms of component. After adaptive segmentation of components histograms of visual image, principal colors and their areas are determined. Then through combining the color areas of these component value, image can be divided into several different colors. Finally, according to principle of color similarity, the image is segmented by cluster segmentation method, with these different colors as the clustering center. Experiments on CASIA 1 mobile robot proves that this method can adaptively segment and detect principal areas of the original image, and can meet real time needs of mobile robot for visual signal processing.