Camera based 3-D object tracking

• Main Author: Ru, Shanshan.
• Other Authors: Xie Lihua
• Format: Final Year Project
• Language: English
• Published: 2011
• Subjects: DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision
• Online Access: http://hdl.handle.net/10356/45867
• Institution: Nanyang Technological University

3-D Object tracking is a procedure to generate the trajectory of an object over time by locating its position in consecutive frames of the online or offline video and at the same time extract its depth information. It is a sub-category of vision based motion analysis. There are numerous methods to solve for object localization problem, mainly based on object features. Compared with 2-D object tracking, besides object localization, 3-D projection needs one more step to reproject the 2-D pixels to 3-D real world space. 3-D object tracking mostly requires at least two cameras to model the binocular vision mechanism and further to retrieve the depth information. In this paper, one color based probabilistic tracking method is used for object localization known as CamShift. In addition, for 3-D reprojection, one set of stereo camera with disparity extraction interface is utilized. The object tracker designed can work in both online and offline modes. The tracker is a primitive model of a computerized arm tracking system. The intention of computerized arm tracking is to monitor the motion the patients to provide an objective measure of the condition and the progress of patients as they undergo therapy.