A New Process for Selecting the Best Background Representatives based on GMM
Nebili WAFA , Seridi Hamid, Kouahla MOHAMED NADJİB
Journal Title:International Journal of Informatics and Applied Mathematics
Background subtraction is an essential step in the process of monitoring videos. Several works have been proposed to differentiate the background pixels from the foreground pixels. Mixtures of Gaussian (GMM) are among the most popular models for a such problem. However, they suffer from some inconveniences related to the light variations and complex scene. In this paper, we propose an improvement of the GMM by proposing a new technique of ordering the Gaussian distributions in the selection phase of the best representatives of the scene. Our approach replaces the usual ranking of Gaussians according to the value of wk ,t/σt with sorting according to their covariance measure which is calculated between each pixel and each of these Gaussians. the obtained results on the Wallflower dataset has proven the effectiveness of the proposed approach compared to standard GMM.