Research on camera calibration in track and field competition video

Camera self-calibration in video is an important technique of 3D reconstruction in augmented reality of sporting events. According the features of sports, our system has realized 3D reconstruction in the video of track and field competition by applied binocular stereo vision knowledge to the camera self-calibration algorithm. In this paper, SIFT is used for corner detection, RANSAC is for the fundamental matrix. Regard two adjacent images as the same scene taken by binocular cameras, we can obtained camera´s internal and external parameters according to the geometrical relationship between the two images in the same scene. Experiments show this calibration system can gain better effect in some scene.