Konferenzbeitrag
First order multiple hypothesis testing for the global nearest neighbor data correlation approach
Lade...
Volltext URI
Dokumententyp
Text/Conference Paper
Dateien
Zusatzinformation
Datum
2010
Autor:innen
Zeitschriftentitel
ISSN der Zeitschrift
Bandtitel
Verlag
Gesellschaft für Informatik e.V.
Zusammenfassung
The growing necessity in multiple targets tracking (MTT) in surveillance systems, with the recent dramatic increase in computational capabilities, has lead to a major interest in improving the performance of classical methods, such as the Global Nearest Neighbor (GNN), to enhanced schemes of Data Correlation. Today, the Multiple Hypothesis Testing (MHT) is generally accepted as the preferred approach for MTT systems, as it demonstrates better results in more complicated and uncertain environments. However, embedding such a mechanism to a deployed GNN-based system requires an extensive software change, and may introduce a major engineering risk to the working environment. Moreover, in a system that is deployed at different sites, which addresses operational environments of different complexities, such a change may be too costly and even superfluous. In this paper we will present a method which will address the challenge of multiple targets tracking in changing environments through a First Order Multiple Hypothesis Testing for a Global Nearest Neighbor engine. We will start with presenting the basics of multiple targets tracking, followed by a review of the proposed solution and conclude with simulations to verify its performance in different scenarios.