Title :
Regularized online Mixture of Gaussians for background subtraction
Author :
Wang, Huifang ; Miller, Paul
Author_Institution :
Centre for Secure Inf. Technol. (CSIT), Queen´s Univ. of Belfast, Belfast, UK
fDate :
Aug. 30 2011-Sept. 2 2011
Abstract :
Mixture of Gaussians (MoG) modelling [13] is a popular approach to background subtraction in video sequences. Although the algorithm shows good empirical performance, it lacks theoretical justification. In this paper, we give a justification for it from an online stochastic expectation maximization (EM) viewpoint and extend it to a general framework of regularized online classification EM for MoG with guaranteed convergence. By choosing a special regularization function, l1 norm, we derived a new set of updating equations for l1 regularized online MoG. It is shown empirically that l1 regularized online MoG converge faster than the original online MoG.
Keywords :
Gaussian processes; expectation-maximisation algorithm; image classification; image sequences; video signal processing; background subtraction; online stochastic expectation maximization viewpoint; regularized online classification; regularized online mixture of Gaussians modelling; special regularization function; video sequences; Approximation algorithms; Approximation methods; Convergence; Cost function; Equations; Mathematical model; Stochastic processes;
Conference_Titel :
Advanced Video and Signal-Based Surveillance (AVSS), 2011 8th IEEE International Conference on
Conference_Location :
Klagenfurt
Print_ISBN :
978-1-4577-0844-2
Electronic_ISBN :
978-1-4577-0843-5
DOI :
10.1109/AVSS.2011.6027331
