Using optical flow for step size initialisation in hand tracking by stochastic optimisation

Authors:
Desmond Chik
Affiliations:
National ICT Australia, Canberra Research Laboratory and Australian National University, Canberra, Australia
Venue:
VisHCI '06 Proceedings of the HCSNet workshop on Use of vision in human-computer interaction - Volume 56
Year:
2006

Citing 6
Cited 0

Pose space deformation: a unified approach to shape interpolation and skeleton-driven deformation

Proceedings of the 27th annual conference on Computer graphics and interactive techniques
Free-viewpoint video of human actors

ACM SIGGRAPH 2003 Papers
Representing the finger-only topology for hand shape recognition

Machine Graphics & Vision International Journal
Automatic Recognition of Colloquial Australian Sign Language

WACV-MOTION '05 Proceedings of the IEEE Workshop on Motion and Video Computing (WACV/MOTION'05) - Volume 2 - Volume 02
Combining 3D flow fields with silhouette-based human motion capture for immersive video

Graphical Models - Special issue on pacific graphics 2003
Full Body Tracking from Multiple Views Using Stochastic Sampling

CVPR '05 Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05) - Volume 2 - Volume 02

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper we use optical flow to initialise step sizes in a stochastic optimisation setting for hand tracking. One can appreciate that most complex hand gestures result in different segments of the hand moving at different speeds. We show that by reflecting this motion difference in our step size initialisation process, the tracking performance can be improved. Significant improvement to tracking accuracy has been observed for gripping motions of the hand.