We propose a Gaussian mixture model with fixed but unknown number of components for background subtraction in infrared imagery. Following a Bayesian approach, our method automatically estimates the number of components as well as their parameters, while simultaneously it avoids over/under fitting. The equations for estimating model parameters are analytically derived and thus our method does not require any sampling algorithm that is computationally and memory inefficient. The pixel density estimate is followed by an efficient and highly accurate updating mechanism, which permits our system to be automatically adapted to dynamically changing visual conditions. Experimental results and comparisons with other methods indicate the high potential of the proposed method while keeping computational cost suitable for real-time applications.
CITATION STYLE
Makantasis, K., Doulamis, A., & Loupos, K. (2015). Variational inference for background subtraction in infrared imagery. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 9474, pp. 693–705). Springer Verlag. https://doi.org/10.1007/978-3-319-27857-5_62
Mendeley helps you to discover research relevant for your work.