Improvement and Evaluation of
Scenario-Type Hypothesis Object Tracking
-
1
Osaka University, Graduate School of Information Science and Technology, Japan
-
2
Osaka University, Cyberscience Center, Japan
Object tracking is one of the major applications of sensor networks. We have proposed an object tracking method called Scenario-type Hypothesis Tracking (SHT) [1]. In our method, a monitoring region is divided into multiple micro cells, and sensors are placed at borders among micro cells (hereinafter referred to as gates), as shown in Fig. 1(a). A sensor is assumed to be able to detect objects and their movement directions. For example, the sensor is realized by putting pair of binary sensors. Sensor information are gathered to a tracking server through wireless or wired networks. The tracking server maintains virtual micro cells and virtual objects which indicate a variety of hypotheses of object trajectory estimation. A virtual object is generated, moved, and removed in virtual micro cells appropriately according to sensor information. By using SHT, it is possible to handle a variety of scenarios including behavior of objects and structure of micro cells, for example, slowdown of objects’ velocities when objects cross each other in narrow passageway, and extension of travel distance of objects in a micro cell due to obstacles. However, the number of virtual objects exponentially increases against parameters, structure of micro cells, and arrival rate of objects [1]. In addition, we should take into account loss of sensor information due to packet loss, failure of sensing, and functional limitation of sensors.
In this article, we propose two improvement methods for SHT. In the first improved method, when predicted sensor information is not obtained at the tracking server, it generates virtual objects for all possibilities of the object’s trajectories to handle loss of sensor information. In the second improved method, new virtual objects are generated from selected virtual objects which have high likelihood, to decrease the number of generated virtual objects while keeping object tracking accuracy. Hereinafter, due to space limitation, we only show one of preliminary simulation results. In the simulation, a monitoring region is composed of 2x2 square micro cells where each micro cell has four gates for each edge. Objects arrive to randomly selected gate following Poisson arrival process, pass through the monitoring region by randomly selecting gates, and depart from the monitoring region. Among intra-micro cells, sensor information is dropped with some probability. Fig. 1(b) shows the tracking accuracy against information loss probability, where the tracking accuracy is the ratio of the number of successful trajectory estimation to the number of objects. Fig. 1(c) shows the virtual objects generation ratio, which means that the ratio of the number of generated virtual objects to the number of arrival objects. In these figures, improved SHT-1 means SHT with the first improved method, and improved SHT-2 means SHT with both improved methods. As shown in Fig. 1(b), it is shown that our improved methods achieve higher tracking accuracy in comparison with SHT regardless of information loss probability. In addition, Fig. 1(c) shows that the number of generated virtual objects in improved SHT-2 is reduced in comparison with improved SHT-1.
This work was partly supported by the KAKENHI (2170075, 1804950) of MEXT in Japan.
References
1. Murata, M., Taniguchi, Y., Hasegawa, G., Nakano, H., An Object Tracking Method based on Scenario-Type Hypothesis Tracking in Segmented Multiple Regions, submitted to ICNS 2010, March 2010.
Conference:
2nd International Workshop on Sensor Networks and Ambient Intelligence In conjunction with PDCAT'09, Hiroshima, Japan, 8 Dec - 11 Dec, 2009.
Presentation Type:
Poster Presentation
Topic:
Poster Presentations
Citation:
Murata
M,
Taniguchi
Y,
Hasegawa
G and
Nakano
H
(2009). Improvement and Evaluation of
Scenario-Type Hypothesis Object Tracking.
Front. Neuroinform.
Conference Abstract:
2nd International Workshop on Sensor Networks and Ambient Intelligence In conjunction with PDCAT'09.
doi: 10.3389/conf.neuro.11.2009.16.015
Copyright:
The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers.
They are made available through the Frontiers publishing platform as a service to conference organizers and presenters.
The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated.
Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed.
For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions.
Received:
24 Nov 2009;
Published Online:
24 Nov 2009.
*
Correspondence:
Masakazu Murata, Osaka University, Graduate School of Information Science and Technology, Osaka, Japan, m-murata@ist.osaka-u.ac.jp