BROWSE
Detail
Conference Paper
Persistent Real-time Camera Tracking
Cited - time in 
Share 
Share
- Authors
- Hye-Mi Kim, Jae-Hean Kim, Jung-Jae Yu, Byoung-Tae Choi
- Issue Date
- 2008-01
- Citation
- Korea-Japan Joint Workshop on Frontiers of Computer Vision (FCV) 2008, pp.59-63
- Language
- English
- Type
- Conference Paper
- Abstract
- Real-time visual tracking has the potential to turn inexpensive cameras into powerful pose sensors for applications. To track the camera path in real-time, Extended Kalman Filter (EKF) is usually used. However, there is an unavoidable limitation to compose state vector in Kalman filter. That is the continuous increase of state vector dimension in Kalman filter. To solve this problem, we proposed the cost function of determining feature component which can be deleted from state vector. Experimental result shows that our method works well to persist camera tracking in real-time.
- KSP Keywords
- Cost Function, Experimental Result, Extended kalman fiLTEr, Feature component, Real-time visual tracking, State vector, camera tracking