Video-Based Motion Capturing for Skeleton-Based 3D Models

In this paper, a semi-automatic method to capture motion data from a single-camera video is proposed. The input video is first segmented and analyzed, and a 3D character model with skeleton rigged is used as a reference model. Then, the reference model is modified to fit the subject’s contour in the starting frame, and the body’s and limbs’ contours of the subject are also specified by the user. Our system then extracts the motion from the video by estimating the reference model’s poses automatically in each video frame forwardly. Finally, the user can help to refine the result through a friendly user interface. Key words: video-based motion capture, user-aid, reference model

• Type: Conference
• Year: 2009
• Where: PSIVT
• Authors: Liang-Yu Shih, Bing-Yu Chen, Ja-Ling Wu
• Keywords: 3D Character Model, Multimedia, PSIVT 2009, Reference Model, Video