Introduction: Microsoft Kinect V2® (Kinect) is a peripheral device of Xbox® and acquires information such as depth, posture, and skeleton definition. In this study, we investigated whether Kinect can be used for human gait analysis.
Methods: Ten healthy volunteers walked 20 trials, and each walk was recorded by a Kinect and infrared- and marker-based-motion capture system. Pearson's correlation and overall agreement with a method of meta-analysis of Pearson's correlation coefficient were used to assess the reliability of each parameter, including gait velocity, gait cycle time, step length, hip and knee joint angle, ground contact time of foot, and max ankle velocity. Hip and knee angles in one gait cycle were calculated in Kinect and motion capture groups.
Results: The coefficients of correlation for gait velocity (r = 0.92), step length (r = 0.81) were regarded as strong reliability. Gait cycle time (r = 0.65), minimum flexion angle of hip joint (r = 0.68) were regarded as moderate reliability. The maximum flexion angle of the hip joint (r = 0.43) and maximum flexion angle of the knee joint (r = 0.54) were regarded as fair reliability. Minimum flexion angle of knee joint (r = 0.23), ground contact time of foot (r = 0.23), and maximum ankle velocity (r = 0.22) were regarded as poor reliability. The method of meta-analysis revealed that participants with small hip and knee flexion angles tended to have poor correlations in maximum flexion angle of hip and knee joints. Similar trajectories of hip and knee angles were observed in Kinect and motion capture groups.
Conclusions: Our results strongly suggest that Kinect could be a reliable device for evaluating gait parameters, including gait velocity, gait cycle time, step length, minimum flexion angle of the hip joint, and maximum flexion angle of the knee joint.
Keywords: Gait analysis; Kinect V2; Motion capture system; Posture recognition; Skeleton definition.
© The Authors, published by EDP Sciences, 2022.