Spatiotemporal Gait Measurement With a Side-View Depth Sensor Using Human Joint Proposals

We propose a method for calculating standard spatiotemporal gait parameters from individual human joints with side-view depth sensor. Clinical walking trials were measured concurrently by Kinect and pressure-sensitive walkway, the Zeno Walkway. Multiple joint proposals generated images stochastic predictor based on algorithm. The are represented as vertices in weighted graph, where weights depend expected lengths between body parts. A shortest path through graph is set of head to foot. Accurate foot positions selected comparing pairs paths. Stance phases feet detected examining motion over time. stance used calculate four parameters: stride length, step width, percentage. constant frame rate was assumed calculation percentage because time stamps not captured during experiment. Gait 52 compared ground truth walkway using Bland-Altman analysis intraclass correlation coefficients. large spatial had strongest agreements (ICC(2, 1) = 1.00 0.98 length normal pace, respectively). presented system directly calculates while previous systems relied indirect measures. Using allows tracking both directions one camera, extending range which subject field view.