Abstract: In the robot soccer system, the environmental information around the robot is characterized in colors. On the analysis of the color threshold segmentation, an improved algorithm practicable in different color environments is proposed to realize the identification of objects in different colors in the field through omni-direction vision and front monocular vision, by the use of which, the robot’s object location through the two visions can be realized and then a more accurate object location can be achieved by combining the information from the two visions through Kalman filter algorithm. Experimental results prove the feasibility and validity of this algorithm which improves the accuracy of medium-and-long-distance object location.

Key words: digital image processing, Kalman filter, localization, color threshold