Vision Processing System Design Of Robot Soccer With Omni-Directional Vision System

This paper focuses on the application of omni-directional vision system(ODVS) with mobile robot being a platform and Robocup middle league being the research topic background. This paper employs Nubot omni-directional mirror and a digital camera to set up a vision system.In the processing of the images, YUV color model, insensitive to light, is employed to calibrate the colors. In the image segmentation period, k-means clustering method is used, which calibrate the thresholds automatically and accomplish the thresholds segmentation in YUV color space with a better result. Then, run-length coding(RLC) is applied to conduct the region segmentation, thus recognizing the ball,goals,field lines according to color and structure of the environment model.For the problem of self-localization, this paper firstly achieves some improvement in randomized hough transform(RHT) to detected lines while obtaining the line parameters. After that, a localization model is established and result is figured out.