Gait disorder detection with M.S. Kinect sensor

In this study, detection of the gait disorders is aimed by using a Kinect sensor along with a key-frame based approach. Gait is evaluated as two major phases. The phases that's been evaluated in this study are stance and swing phases. In order to determine the gait phase, maximum distance value of the ankle relative to hip center joint is used. Frames are labeled with the gait phase of each leg of the bodies inside the view. Afterwards kinematic features of each leg according to the phases they are in are calculated. The calculated kinematic features are used as the input to the finite state automaton that's been modeled for each gait disorder. By the help of finite state automaton, at the end of the transitions of kinematic values inside the spatio-temporal frames, gait disorders are recognized. Eventually this study introduces that the detection of over-ground gait cycles and the detection of the gait disorders, if there is any kinematic anomaly that characterizes a known gait disorder during the gait cycles, are possible by the use of an inexpensive and portable sensor