Towards Automatic Gait Assessment by Means of RGB-D Mocap

The present paper describes the research done to empirically validate the use of RGB-D cameras for orthopedic rehabilitation purposes. In order to assess the outcome of optical sensors a set of new generation inertial sensors have been adopted as well. Both kind of sensors have been used to track the same scene simultaneously in order to have comparable results. The aim of the work is provide evidence of performances of the newly introduced sensors (optical and inertial) and to automate acquisition and data elaboration in the medical field. The extraction of the only meaningful data from a gait, for instance, gives the change to perform an automatic preliminary analysis on normal people as well as on patients with known pathologies or disorders. In particular, this could allow determining without human intervention any standard disorders, according to literature classifications, eventually affecting the gait. The automatic procedure from acquisition to diagnosis constitutes a challenging topic of research and a success in terms of reliability, usability and physiological acceptance by technicians would radically impact medical practices in orthopedic rehabilitation centers.