Model-based automatic tracking of articulated human movement

This study applied a vision-based tracking approach to the analysis of articulated, three-dimensional (3D) whole-body human movements. A 3D computer graphics model of the human body was constructed from ellipsoid solids and customized to two gymnasts for size and colour. The model was used in the generation of model images from multiple camera views with simulated environments based on measurements taken on each of three synchronized video cameras and the lighting sources present in the original recording environment. A hierarchical procedure was used whereby the torso was tracked initially to establish whole-body position and orientation and subsequently body segments were added successively to the model to establish body configuration. An iterative procedure was used at each stage to optimize each new set of variables using a score based on the RGB colour difference between the model images and video images at each stage. Tracking experiments were carried out on movement sequences using both synthetic and video image data. Promising qualitative results were obtained with consistent model matching in all sequences, including sequences involving whole-body rotational movements. Accurate tracking results were obtained for the synthetic image sequences. Automatic tracking results for the video images were also compared with kinematic estimates obtained via manual digitization and favourable comparisons were obtained. It is concluded that with further development this model-based approach using colour matching should provide the basis of a robust and accurate tracking system applicable to data collection for biomechanics studies.