Recovery of 3D human posture based on monocular vision technology

Abstract

This paper presents a novel approach of 3D human posture recovering from monocular vision frame under the condition of which at least one depth value of a joint point is known in each monocular vision frame. This approach simplifies the calibration of CCD camera, and realizes the recovery of 3D human posture from monocular vision. Moreover, the methodology based on monocular vision can extensively releases from the space and illumination restricts and overcome the greatest shortcoming on matching corresponding characteristics in multi-vision system. The procedure of 3D human posture recovering is demonstrated in this paper for proving the feasibility of the approach. ©2009 IEEE.