摘要：Visual background extractor (Vibe), a universal background subtraction algorithm, which has the characteristics of fast operation, high accuracy and wide application range. But there are also difficulties in eliminating the "ghost" and adapting to the light mutation, in this paper, an improved Vibe algorithm is proposed. It draws on the W4 theory and the mean value method in establishing the background model to suppress ghosts and some shadows, the new background model of Vibe was created by adding light mutation criterion to the motion detection algorithm, so that it could adapt to the light mutation rapidly, the experimental results show that the proposed method have good performance in removing interfaces as ghost shadow, when any sudden light change of circumstance occurs, the algorithm can also detect and update pixels statistical parameters adaptively
