Dr Andreas Aristidou
|
Office Location: BN3-03 (Signal Processing South Lab) Position: Researcher |
|
Real-Time Hand Pose Tracking using Optical Motion Capture
In recent years there has been a growing demand for reliable hand motion tracking systems, a technology used to turn the observations of a moving hand into 3D position and orientation information. However, building a fast and effective hand pose tracker remains challenging; the high dimensionality of the pose space, the ambiguities due to self-occlusions and the significant appearance variations due to shading make efficient tracking difficult. Marker-based motion capture has been demonstrated in several interactive systems (including but not limited to hand interaction); the results are highly accurate and easy to configure. There are, however, instances where we do not have many markers available or it is impossible to attach 3 markers on each limb segment; the large number of markers needed is often prohibitive. It may therefore be infeasible to reproduce the tracked object animation and reconstruct its skeleton model (i.e. the hand model). Hence, it is necessary to find a new way of capturing the movement of these articulated models, using the minimum possible number of markers. Thus, instead of attaching 3 markers on each limb segment, we investigate a system in which a single marker is attached and captured on each finger (end effector), 1 marker at the chain base (root) and 2 markers at strategic positions to help us define the hand orientation. The markers' positions are tracked using an optical motion capture system, such as Phasespace. However, prior knowledge about the geometry of the hand, the hand model and the restrictions of each joint are required. Joint constraints are applied to ensure that finger motion is within a feasible set, giving a visually natural motion of the hand. An Inverse Kinematics solver (FABRIK) is incorporated to estimate the remaining joint positions and to fit them to the hand model. Physiological constraints related to the hand anatomy are then incorporated to restrict the motion only to natural possible poses, without violating any model constraint. Finally, a mesh deformation algorithm has been applied to drive the animation of the underlying hand skeleton using a set of per-bone weights. The implemented mesh videos are compared with their true hand motions and the results verify that the suggested method is effective and real-time implementable.
Related Publications:
[2] |
Andreas Aristidou, Joan Lasenby, "Inverse Kinematics solutions using Conformal Geometric Algebra", In L. Dorst and J. Lasenby (Eds), Guide to Geometric Algebra in Practice, Springer Verlag, 2011. |
[1] |
Andreas Aristidou, Joan Lasenby, "Motion Capture with Constrained Inverse Kinematics for Real-Time Hand Tracking", In IEEE Proceedings of the 4th International Symposium on Communications, Control and Signal Processing (ISCCSP'10), Cyprus, May 3-5, 2010. |
