The use of kernel methods for audio events detection

Author

Nasser, Alissar ; Hamad, Denis ; Rouas, Jean-Luc ; Ambellouis, Sébastien

Author_Institution

LASL/ULCO, Calais

fYear

2008

Firstpage

1

Lastpage

6

Abstract

This paper presents an approach for an automatic surveillance system in public transport by analyzing audio signals recorded in the vehicle in order to detect several abnormal behaviors. We try to visualize audio signals by projection methods like PCA and kernel PCA. We use also unsupervised classification (clustering) methods to separate the audio signals into their components precisely we are using K- means and kernel K-means.

Keywords

audio signal processing; pattern clustering; signal classification; signal detection; source separation; surveillance; abnormal behavior detection; audio event detection; audio signal separation; audio signal visualization; audio signals analysis; automatic surveillance system; kernel k-means method; public transport; unsupervised classification method; unsupervised clustering method; Clustering algorithms; Data mining; Data visualization; Event detection; Feature extraction; Kernel; Principal component analysis; Signal analysis; Surveillance; Vehicles; Kernel methods; MFCC; audio detection; clustering; projection;

fLanguage

English

Publisher

ieee

Conference_Titel

Information and Communication Technologies: From Theory to Applications, 2008. ICTTA 2008. 3rd International Conference on

Conference_Location

Damascus

Print_ISBN

978-1-4244-1751-3

Electronic_ISBN

978-1-4244-1752-0

Type

conf

DOI

10.1109/ICTTA.2008.4529996

Filename

4529996