WSEAS Transactions on Information Science and Applications
Print ISSN: 1790-0832, E-ISSN: 2224-3402
Volume 10, 2013
A Novel Algorithm for Source Localization Based on NonnegativeMatrix Factorization using αβ -Divergence in Cochleagram
Authors: ,
Abstract: In this paper, a localization framework based on a nonnegative matrix factorization using a family of αβ-divergence and cochleagram representation is introduced. This method provides accurate localization performance under very adverse acoustic conditions. The system consists of a three-stage analysis,the first stage: the source separation using NMF based on αβ-divergence where the decomposition performed in cochleagram domain. In the second stage the estimated mixing matrix used to estimate the Time Difference of arrival (TDOA). Finally the Time Difference of Arrival estimates can be exploited to localize the sources individually using the Scaled Conjugate Gradient algorithm (SCG) ,where SCG has advanced compared to other conjugated gradient algorithms. Experiments in real and simulated environments with different microphone setups in 2-D plane and 3-D space, are discussed, showing the validity of the proposed approach and comparing its performance with other state-of-the-art methods.
Search Articles
Keywords: Blind source separation (BSS), Nonnegative Matrix Factorization (NMF), αβ divergence, sound source localization (SSL)