Improved color object recognition method of humanoid soccer robot
一种改进的类人足球机器人彩色目标识别算法

As the vision demands of humanoid soccer robot, a color object recognition method was proposed that combined region growing with the adaptive threshold update algorithm based on shape discrimination. The method divided the image into the high saturation region and the low saturation region by the S component in the HSI color space. The region growing was used to recognize the object by H component in the high saturation region. The threshold was adaptively updated by shape discrimination and was used in region growing in order to accurately recognize the color object stably. The successful applications of the method on a real humanoid soccer robot show that the method has well adaptable and robust for changing illumination because color objects are recognized stably under different illumination condition. So the method has good recognition effects.