Gait symmetry measurement method based on a single camera

Abstract

Changes in gait symmetry may indicate health deterioration of seniors, such as inefficiency and loss of balance control. Measuring symmetry of seniors’ daily gaits can provide essential information on health status and functional abilities of the seniors, and therefore has attracted increasing attention in the field of elderly health care. Current gait symmetry measurement using a single camera is usually restricted to walking directions, because most studies require subjects to walk perpendicularly to the camera’s optical axis. This limits the application of single-camera-based systems in daily-life gait measurement. In this paper, we propose a gait symmetry measurement method using a single camera, unconstrained by walking directions. We first establish mathematical relationship between projections of step length and body’s height on the image plane in three consecutive steps, and then utilize the mathematical relations to generate a formula for symmetry ratio of lengths of two adjacent steps, and meanwhile eliminate impacts of the walking directions. For a monocular monitoring video, the proposed method segments the sequence of human silhouettes into multiple steps after detecting a moving human, and then makes use of information of these steps to compute the symmetry ratio according to the derived formula, and finally provides gait symmetry measurement irrespective of walking directions. Experimental results indicate that our method can achieve accurate gait symmetry measurement, in good agreement with actual measured data, in unconstrained environments where subjects’ walking directions are not strictly restricted.