Object Recognization And Localization Based On Omni-Vision

Robotic soccer has become a popular research domain of robotics and artificial intelligence. In various soccer robots, autonomous soccer robot has big values of application, it's research can be applied to autonomous driving, guide robot, unknown terrain exploring, nuclear power plant robot, product transmit robot, etc.. We have made a research on Omni-vision System, which is used by autonomous soccer robot, and Omni-vision System is involved in Image Processing, Image Analysis, Image Understanding, etc.. This paper described the hardware of Omni-vision System, 4 parts of the software in Omni-vision System: video capture, Image Pre-processing, object recognition based on color information and object localization based on Omni-vision. Then we described the procedure of designing and developing Omni-vision System mounted in autonomous soccer robot.The hardware of Omni-vision System includes a reflector which is reelingly symmetric, a camera, a support glass and a support bracket. The most important trait of this vision system is that its view field is 360o horizontally and some degree vertically. So this system is fast in Object Tracking, and vast in view field, and it can adapts to all games.The Object Recognition and Localization System Based on Omni-vision is the core of autonomous soccer robot, it is important to the performance of the whole robot system. Object Recognition based on color information and Object Localization based on Omni-vision are the most important parts in Omni-vision system. During Object Recognition, we used color query in Image Segmentation, which made Clustering to every pixel much faster and easier. We did Object Recognition based on priori information, so we could get rid of noise and recognize the real object. During Object Localization, we researched on angle localization and distance localization. In the end, we did Object Localization in soccer robot game based on compound reflector, and we obtained good result. Our experiment proved the validity of the Object Recognition and Localization System Based on Omni-vision.