Creating a "Personalised, Immersive Sports TV Experience" via 3d Reconstruction of Moving Athletes

28 Introduction Over the last years, several methods of enhancement in sports television were introduced, e.g. a moving line enabling the comparison of an athlete’s attempt with the world record, or the overlay of two competitors for comparison of their skiing technique etc. Due to the nature of traditional television, these enhancements were previously limited to 2d sequences the TV viewer cannot interact with. With the advent of MPEG-4, advanced set-top boxes enable the interactive visualization of animated 3d content. However, the creation of suitable content that uses the 3d features of the MPEG-4 format is much more difficult than the production of ordinary TV content, particularly in the case of 3d content representing real world events. In order to bridge the gap between technical possibilities of MPEG-4 and available tools for creating high quality content, PISTE aims at automatically converting ordinary images from TV cameras to a 3d scene description which contains an animated body model of the athlete in its 3d environment with accurate body movements. These 3d animations enable several novel viewing modalities: – The TV viewer interactively specifies position and direction of the camera while watching the sports event. – Several athletes can be watched simultaneously within the same environment for comparison. – By overlaying a metric grid, the athlete’s attempt can be analysed in detail. Creating a »Personalised, Immersive Sports TV Experience« via 3d Reconstruction of Moving Athletes

[1]  Rolf Adams,et al.  Seeded Region Growing , 1994, IEEE Trans. Pattern Anal. Mach. Intell..

[2]  O. Faugeras Three-dimensional computer vision: a geometric viewpoint , 1993 .

[3]  Georgios Tziritas,et al.  Bayesian Level Sets for Image Segmentation , 2002, J. Vis. Commun. Image Represent..

[4]  Henry. Dreyfuss,et al.  The measure of man , 1960 .

[5]  Krzysztof Walczak,et al.  X-VRML-XML based modeling of virtual reality , 2002, Proceedings 2002 Symposium on Applications and the Internet (SAINT 2002).

[6]  Andrew W. Fitzgibbon,et al.  Bundle Adjustment - A Modern Synthesis , 1999, Workshop on Vision Algorithms.

[7]  Seth J. Teller,et al.  Acquisition of a large pose-mosaic dataset , 1998, Proceedings. 1998 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (Cat. No.98CB36231).

[8]  Christian Bouville,et al.  Enhanced sports broadcasting by means of augmented reality in MPEG-4 , 2001 .

[9]  F. A. Seiler,et al.  Numerical Recipes in C: The Art of Scientific Computing , 1989 .

[10]  David C. Hogg,et al.  Towards 3D hand tracking using a deformable model , 1996, Proceedings of the Second International Conference on Automatic Face and Gesture Recognition.

[11]  Timothy F. Cootes,et al.  Training Models of Shape from Sets of Examples , 1992, BMVC.