Y.Y. Wai, T. Lee, and W.K. Cham (PRC)
Motion capture, monocular image sequence, depth extraction, genetic algorithm, non-coplanar.
The problem of motion capture from an image sequence taken from a video camera is addressed in this paper. A key step of the motion capture is to determine the distance of an object from the image plane since the images only consist of 2D information. With this limitation, only co planar motion can be captured with good results. We show that if the actual size of the object is known or can be estimated, good estimation of the motion can be obtained by assuming smooth motion transition between image frames. This is achieved by formulating the depth extraction process as an optimization problem, and we propose to solve this problem using genetic algorithm. The experiment results show that our solution can extract non-coplanar motions.
Important Links:
Go Back
