Classification of Vehicles in a Multi-Object Scenario using Acoustic Sensor Arrays
2008 (English)Article in journal (Refereed) Submitted
The organization of distributed sensor networks is a topic gaining much interest. Detection, tracking and classification of objects moving within a geographic area is one possible application for such network. This article consider the classification of vehicles moving in an area monitored by a network of acoustic sensors. It is studied how to classify a target vehicle from acoustic recordings in which also interfering vehicles are present. Classification is made using a support vector classifer. The focus of this paper is on a combined beam-former and independent component analysis procedure used to separate the acoustic signature of a target vehicle from signatures of present interferences. An evaluation of the method is made on real data. The results show that the source separation method can help to improve classification performance. Properties of the joint separation-classification technique is analyzed and it is discussed under which circumstances the separation is most effcient.
Place, publisher, year, edition, pages
Support Vector Classifcation, Beamforming, Independent Component Analysis, multi object classifcation, distributed sensor network
Atom and Molecular Physics and Optics
IdentifiersURN: urn:nbn:se:liu:diva-14878OAI: oai:DiVA.org:liu-14878DiVA: diva2:25481