3-D Facial Motion Estimation Using Iterative Extended Kalman Filter


The KIPS Transactions:PartB , Vol. 8, No. 1, pp. 28-34, Feb. 2001
10.3745/KIPSTB.2001.8.1.28,   PDF Download:

Abstract

Accurate 3-D facial motion estimation using computer vision is required for 3-D view control in desktop VR system or simulator and gaze detection on a monitor, etc and it has recently been researched vigorously. Most previous researches use a EKF (Extended Kalman Filter) for 3-D facial motion estimation, but it has restriction of choosing accurate initial filter parameters. In addition, the EKF cannot estimate 3-D facial motion when a user changes his face direction abruptly. To overcome such problems, we use IEKF (Iterative Extended Kalman Filter) transformed from EKF. IEKF is that when the estimated error covariance value calculated form IEKF exceeds in a predetermined threshold, it adjusts the initial parameters(rotational and translational velocities or accelerations) adaptively so that the 3-D facial motion can be estimated exactly. As experimental results, IEKF can even estimate 3-D facial motion accurately in case of abrupt facial directional change.


Statistics
Show / Hide Statistics

Statistics (Cumulative Counts from September 1st, 2017)
Multiple requests among the same browser session are counted as one view.
If you mouse over a chart, the values of data points will be shown.


Cite this article
[IEEE Style]
K. R. Park and J. H. Kim, "3-D Facial Motion Estimation Using Iterative Extended Kalman Filter," The KIPS Transactions:PartB , vol. 8, no. 1, pp. 28-34, 2001. DOI: 10.3745/KIPSTB.2001.8.1.28.

[ACM Style]
Kang Ryoung Park and Jai Hie Kim. 2001. 3-D Facial Motion Estimation Using Iterative Extended Kalman Filter. The KIPS Transactions:PartB , 8, 1, (2001), 28-34. DOI: 10.3745/KIPSTB.2001.8.1.28.