A subpixel algorithm for the reconstruction of kinematics from video sequences in orthostatic postural tests

A marker-free method for the reconstruction of kinematics from video sequences during posturographic tests is presented. The image processing algorithm relies on a two-phase motion estimation: a block matching technique, and a fine interpolation that was verified as granting a subpixel resolution. The algorithm was previously evaluated in terms of robustness with respect to SNR, and then applied to the recognition of postural strategies from real sequences during stabilometric trials. The obtained results confirm the validity of the method and thus favours the exploitation of video signal processing for the determination of kinematics in biomechanical tasks.