Embedded realtime feature fusion based on ANN, SVM and NBC

Author

Starzacher, Andreas ; Rinner, Bernhard

Author_Institution

Inst. of Networked & Embedded Syst., Klagenfurt Univ., Klagenfurt, Austria

fYear

2009

fDate

6-9 July 2009

Firstpage

482

Lastpage

489

Abstract

Artificial neural networks (ANNs), support vector machines (SVMs) and naive Bayes classifiers (NBCs) are common tools for multisensor data fusion applications. In this paper ANN, SVM and NBC are applied to embedded realtime feature fusion and compared to different algorithms concerning classification execution time as well as classification rate. These algorithms are implemented on our three-layered multisensor data fusion architecture and applied to traffic monitoring where we are focusing on fusing data originating from distributed acoustic, image and laser sensors for vehicle classification and tracking. The evaluation of the algorithms is performed on our embedded platform and has shown promising results concerning realtime classification execution time and classification rate.

Keywords

Bayes methods; embedded systems; neural nets; pattern classification; sensor fusion; support vector machines; acoustic sensor; artificial neural network; embedded realtime feature fusion; image sensor; laser sensor; multisensor data fusion; naive Bayes classifier; support vector machine; traffic monitoring; vehicle classification; vehicle tracking; Acoustic sensors; Artificial neural networks; Classification algorithms; Image sensors; Laser fusion; Monitoring; Niobium compounds; Sensor fusion; Support vector machine classification; Support vector machines; Realtime feature fusion; embedded system; naive Bayes; neural network; support vector machine;

fLanguage

English

Publisher

ieee

Conference_Titel

Information Fusion, 2009. FUSION '09. 12th International Conference on

Conference_Location

Seattle, WA

Print_ISBN

978-0-9824-4380-4

Type

conf

Filename

5203644