Video Foreground Detection in Non-static Background Using Multi-dimensional Color Space

Detecting foreground (FG) and suppressing background (BG) is a vital task in video sequence analysis. This task is very challenging when the BG is non-static. Although there have been many algorithms proposed in literature, most of them are complex in terms of either mathematical modeling or computational requirements. In this paper, we experiment two simple algorithms for video FG detection using multi-dimensional color space when the BG is non-static. The algorithms utilize pixel level temporal intensity for FG and BG classification. The algorithms are tested on two sets of outdoor video sequences where the backgrounds are non-static. The experiment results show that the algorithms adequately perform well on the given environments.