In this paper an algorithm of moving objects detection in video image was proposed

which is based on the background codebook model. In this algorithm

normalized Mann-Whitney rank-sum test was used to adjust the detection threshold of codebook model

and Mean Shift was also used to update the code and variance. So this algorithm can achieve the updating of codebook model and the detection of moving objects at the same time. The experimental results show that based on the distribution independence of Mann-Whitney rank-sum test this method can detect the moving objects efficiently under different conditions.