Tracking People with Active Cameras via Bayesian Risk Formulation
-
- Yildiz Alparslan
- Sato Lab., Graduate School of Engineering Science, Osaka University
-
- Takemura Noriko
- Sato Lab., Graduate School of Engineering Science, Osaka University
-
- Iwai Yoshio
- Iwai Lab., Graduate School of Engineering, Tottori University
-
- Sato Kosuke
- Sato Lab., Graduate School of Engineering Science, Osaka University
Search this article
Abstract
In this study, we introduce a system for tracking multiple people using multiple active cameras. Our main objective is to capture as many targets as possible at any time, using a limited number of active cameras. In our context, an active camera is a statically located pan-tilt-zoom camera. The use of active cameras for tracking has not been thoroughly researched, because it is relatively easier to set up and use static cameras. However, there are many properties of active cameras that we can exploit. Our results show that an approximately two-fold increase in relative accuracy can be achieved without any significant increases in computational costs. Our main contributions include removing the necessity for the individual detection of each tracked target, estimating the future states of the system using a simplified fluid simulation, and finally unifying the active camera tracking method using a minimum risk formulation. We also improved the accuracy by developing an efficient method for attracting cameras towards targets located far away from the present camera configuration.
Journal
-
- IEEJ Transactions on Electronics, Information and Systems
-
IEEJ Transactions on Electronics, Information and Systems 134 (6), 870-877, 2014
The Institute of Electrical Engineers of Japan
- Tweet
Details 詳細情報について
-
- CRID
- 1390282679583729792
-
- NII Article ID
- 130004870065
-
- NII Book ID
- AN10065950
-
- ISSN
- 13488155
- 03854221
-
- NDL BIB ID
- 025598374
-
- Text Lang
- en
-
- Data Source
-
- JaLC
- NDL
- Crossref
- CiNii Articles
-
- Abstract License Flag
- Disallowed