The aim of this paper is to propose and initially evaluate the method for comparison, analyses and visualization of similarities and differences between three - dimensional trajectories of body joints in human motion activities. The method is based on dynamic time warping (DTW) motion capture (MoCap) signal matching in hierarchical kinematic model. We have used quaternion-based angles description. To evaluate our novel MoCap evaluation algorithm we have used recordings of black belt Shorin-Ryu karate master and black belt Oyama karate master. Both of them performed mawashi-geri kick with a right leg. The DTW normalized distance (one minus module of dot product of quaternions) for hips joints equals 0.003, for right thighs 0.018 and 0.059 for right legs. The proposed evaluation and visualization technique seems to be a valuable tool for advanced human motion analysis. The DTW signal mapping plots are even more informative when one can rotate them and change number of points to display to see the kinematic progress.
© Copyright 2017 2017 IEEE Symposium Series on Computational Intelligence (SSCI). Published by IEEE. All rights reserved.