Vision system is the critical way to obtain environmental information for humanoid soccer robot. During the competition, because of the variable illumination conditions on the field, the Recognition algorithm using traditional segmentation is not satisfying. An object recognition algorithm, which combined target color with shape information is presented in this paper. This algorithm implements rapid image threshold segmentation based on color information in the HSI color space, obtain the pixel of the object, and joined the adaptive threshold update, then use object pixel and shape information implement optimizing edge detection to recognize the object, finally obtain the accurate target location information in the image. Experiments show that this algorithm can be a long time to obtain the target location information in the image in different illumination conditions, satisfy the requirement of the real-time.