Image segmentation algorithm for humanoid robot color space

This paper presents a novel image segmentation algorithm of improving the quality of color images for humanoid robot soccer game. For getting quickly the color image, hue-saturation-intensity (HSI) color space is used in which an enhanced fuzzy K means clustering method is designed. While designing the fuzzy K means clustering method, all the similar pixels are found by judging the main feature of current pixel according to an intensity (I value) in the related formula. And the hue (H value) or I value is dynamically decided to be used by the distribution of saturation (S value), so that improving the quality of the HSI clolor space model and enhancing the visual capability of the humanoid robot to be adapt to the complex game environment. The experiment results indicate the effectiveness of the proposed method.