A MODEL-BASED 3-D TRACKING OF RIGID OBJECTS FROM A SEQUENCE OF MULTIPLE PERSPECTIVE VIEWS

A method of tracking multiple objects of known geometry using multiple cameras is proposed. Our approach differs from the previous approache s in that the object geometry is tightly integrated into the tracking process. The major contribution is threefold: Firstly, multiple camera s are used to improve the accuracy of the estimated posture parameters . Additional formalism required by considering multiple images is nice ly integrated into the tracking model, and is handled effectively. Sec ondly, the feature tracking is facilitated by integrating the measurem ent and dynamic models into the matching process, thereby improving th e accuracy and robustness of the feature correspondence. Thirdly, ambi guities that may arise in the course of the feature matching are resol ved by the statistical analysis and the visibility test. The entire pr ocess from the image sequence to the posture parameters has been compl etely automated into a single, seamless process, and has been extensiv ely tested on synthetic and real images. (C) 1998 Elsevier Science B.V . All rights reserved.