Low cost vision system for human gait acquisition and characterization

This paper presents a low cost system for human gait analysis. Two opposite-faced web cameras are used to acquire images of the walking of a person carrying a set of passive marks, where the color is chosen to contrast with the ambient dominant color. It is also used a treadmill with passive marks, where the user walks at different speeds. The acquired trajectories of the marks are used to determine body joint angles and other 3D crossed angles, obtained by both opposite sides from video images processing. With this low cost measurement system the analysis and reconstruction of human gait can be done with a mean error of 2 degrees, becoming a good alternative to more expensive systems to be used in human gait characterization. The system can be used to detect human gait pathologies and to accomplish physical rehabilitation.