A GENERAL FRAMEWORK FOR 3D SOCCER BALL ESTIMATION AND TRACKING (TP-P1)

Author(s) : Jinchang Ren (Digital Imaging Research Centre, Kingston University, UK) James Orwell (Digital Imaging Research Centre, Kingston University, UK) Graeme Jones (Digital Imaging Research Centre, Kingston University, UK) Ming Xu (Digital Imaging Research Centre, Kingston University, UK)

Abstract : A general framework for automatic 3D soccer ball estima-tion and tracking from multiple image sequences is pro-posed. Firstly, the ball trajectory is modelled as planar-curves in consecutive virtual vertical planes. These planes are then determined by two ball positions with accurately estimated height, namely critical points, which are ex-tracted by curvature thresholding and nearest distance of 3D lines from single or multiple views respectively. Fi-nally, unreliable or missing ball observations are recov-ered using geometric constraints and polynomial interpo-lation. Experiments on video sequences from different cameras, with over 5000 frames each, have demonstrated a comprehensive solution for accurate and robust 3D ball estimation and tracking, with over 90% ball estimations within 2.5 metres of manually derived ground truth.

Menu