Spatio-temporal features of human gait kinematics based on extended nearest-neighborhood tracking algorithm

An extended nearest-neighborhood (ENN) tracking algorithm was appropriated in designing spatiotemporal features of human gait kinematics under low illumination. Active markers in the form of light-emitting diodes (LEDs) were positioned at anatomical landmarks to measure the coordinated kinematics of human joints. Human gait is analyzed in various perspectives, e.g. front view, left and right side view of the test subjects whose age range from 18-21 years. Front view gait reveals the center of mass of the subject and the side view gives the angular displacement of each joint in the body with respect to its neighboring joints. Image segmentation and binary labeling were also applied on LED images to determine automatically its threshold value. Results have shown that ENN is more robust than typical Kalman filter in predicting LED positions due to obscured view during gait recording. © International Federation for Medical and Biological Engineering 2007