THREE-DIMENSIONAL SPORT MOVEMENT ANALYSIS BY MEANS OF FREE FLOATING TV CAMERAS WITH VARIABLE OPTICS

Abstract

INTRODUCTION: Video analysis and off-line manual digitalization is usually used for 2-D and 3-D studies of human movement in sport science. The main advantage of this approach, with respect to the recourse to opto-electronic automatic motion analyzers, is the high flexibility in system set-up, the avoidance of marking procedures and the possibility of successful operation in a wide range of environmental situations. Such features turn out to be particularly important for recordings to be performed in the frame of high-level competition, when the experimental set-up must be adapted to a pre-defined competitive environment, without interfering with the performances of the athletes. However, when methods proper to conventional close-range photogrammetry are used, most of the advantages offered by the flexibility of video analysis are not obtained. Particularly critical is the restriction of the useful calibrated volume to the field of view made possible by fixed pairs of TV cameras. In this case the useful sequence of images (where the dimension of the acquired subject allows one to limit macroscopic digitalization errors) is often insufficient for the analysis of a complete movement cycle. This limitation hinders a fruitful application of video analysis in the frame of sport activities (alpine and Nordic skiing, swimming, track and field) in which the execution of the particular technical movement is performed within a large physical space. A solution to the problem is proposed based on the use of free moving and zooming cameras. The corresponding dedicated software for repeated calibration based on Direct Linear Transformation (DLT)(Abdel Aziz and Karara, 1971) is described. Results of recording performed in the laboratory are discussed aiming at the validation of the implemented method. The description of the methodology for the recording of sport activities and the presentation of the related results confirm the operational feasibility of the proposed method and the reliability of the resulting quantitative kinematics analysis.