The Research Of Target Recognition And Position Based On Omnidirectional Vision Soccer Robot

Robot vision system has a strong engineering background and involves a broad range of areas. As an important approach of acquiring environmental information, the vision subsystem is critical to robot localization. If a mobile robot wants to move freely in a certain environment, it must know its location in this environment. For mobile robots, the perception approach of using panoramic camera system for external information is perceived in recent years and based on this imaging system, the advantage of acquiring information is the full 360-degree image information. around the robot is available.In this thesis,it is focused on the research of omnidirectional vision system with mobile robot being a platform and implemented a novel omni-vision system for the application in the middle-size league of RoboCup--The World Cup of Soccer Robots.Process the image collected from the omnidirectional sight soccer robots and distinguished the target objects property among it .Otherwise ,position itself and target object according the result of the processed image. In the procedure of distinguish target object, firstly pre-process the collected image is need include image sharpening and image smoothing which the way to enhance the image ;secondly extract the target object form the background (green football arena) before distinguish it . Cut up the image in this thesis under the method which combined by HSI color thresholding and REL Segmentation, this method can split the target object from the background quickly and than combine the same color area into a big one . At last ,based on the distinguish method to distinguish the extracted objects, the author propose the scanning method with the step of circle regions scan the image and recognize the target object by color thresholding.In the proess of soccer robot position ,the method in the thesis based on position method of white lines grid. First , split the whole square into many grids which board is 10cm.We can calculate the grids number between soccer robot and the white line and then the distance is clear based on the transform among robot coordinate and the whole world .And furthermore we can know the minimal distance between soccer robot and target object through calculate the square number between it .And last the whole procedure of soccer robot and target object position are finished.At end of this thesis,the author designed a target recognizing and self-positing experiment and had a satisfied result. The final conclusion has been given which including the innovation points of the research and the main research results, and simultaneously, indicating shortages and problems need for further research.