Human Motion Analysis Based On Video Sequences

Visual analysis of human motion has been receiving increasing attention from researchers in the fields of image processing and computer vision during the past few years. It has a lot of applications in virtual reality, smart surveillance system, advanced user interface, motion analysis and video compressing, etc.The object researched in this thesis is video-based human motion analysis that track, capture and analyze human motion in monocular video. On the basis of sufficient review to the related works, the thesis introduces a system designed and implemented on Video Human Motion Tracking and Capture. The main research contexts and results are as follows:The background subtraction method is used to retrieve the foreground. With gradient modeling, moving cast shadows are eliminated. As a result, whole human body is extracted from monocular video sequences; To eliminate the effect of luminance variance of markers caused by illumination, a robust method based on normalized RGB color which is light-invariant is presented to detect the markers; Geometrical relationship between low limbs and rules of human motion are used to partially solve the problem of self-occlusions, which makes the tracking more continuous; Generate a motion capture data file, which is used to analyze the gait trait. Experiment results are given in the thesis.The research provides a step of progress on the way towards tracking and capture human motion from monocular video sequences. The experimental results demonstrate the feasibility and practicability of our approaches. The research can be applied in computer-aided diagnoses of apoplexy.