Research And Development Of "virtual Tennis" System Based On Computer Vision

Nowadays, computer vision has been widely used in national security, industrial production, transportation and sports, and becomes more and more important. Hawk-eye system, applied in the International Tennis Open, is one classical example which combined computer vision technology with sports, and inject new vitality to this race. In addition, the cause of science and technology museum launched quickly after entering 20th century 90s, how to combine computer vision and science exhibits is proposed as a new issue."Virtual Tennis" System, which is based on computer vision technology and takes Hawk-eye System as technical background, provides an effective platform the above-mentioned combination. In the process of system designation and implementation, we mainly focus on the technology of camera calibration, motion object detection, targets location, binocular stereo vision and so on.The "Virtual Tennis" system based on computer vision is researched and implemented on the basis of reading a great quantity of references. Response to the system requirements, we use DAHENG--SV400FM, which is a kind of high-speed industrial camera, to acquire image. According to the thesis of Tsai Two-step, camera calibration is taken, and camera's internal parameters, external parameters, transformation relationship between two cameras are obtained through this step, and get ready for obtaining target's 3D information. The method of background difference is used to detect moving targets, combined with updating background image in time to improve the accuracy of detection. A new way is proposed, which includes background difference, feature point statistics, mathematical morphology, when locating target. It is not sensitive to light, has better robustness. After obtained the target's 2D center in each image, we use algorithm of binocular vision to get target's 3D information according to the only one target's feature and real-time requirement. For data's error, curve fitting is used to smooth data, and the effect of three-dimensional reconstruction is gotten. It is proved by the results that the method mentioned in this article runs fast, has better robustness, its'precision error is about 1cm, and meets the system's requirement.