Reconstruction of Human Behavior and Movement Feature Recognition Based on the Video Image

Abstract

Based on the analysis of human motion with the development of computer vision technology gradually become the research focus, while intelligent monitoring and other related real needs also contributed to its rapid development, attracting more and more researchers to join them. It includes both limbs motion tracking, as well as tracking the movement of the whole body, but does not include the subtle body movement, such as facial expressions and gestures of the track. The monocular video motion tracking method in this article are used in a variety of models, including the human body geometry, said camera imaging model and motion model used. Geometric model of the human body by the appearance of the model human skeleton model and composition of each joint skeleton model the relative proportions of the length of a value obtained from anthropometric measurements, and the establishment of a local coordinate system at each joint. Camera imaging using fluoroscopy model and a mathematical model based on this request to the relative three-dimensional coordinates of each joint between the sports model based on kinematics knowledge polynomial representation.