Abstract: In order to improve the adaptive ability and update efficiency of the classic codebook background model when detecting moving targets in complex dynamic scene, an improved codebook model algorithm is proposed. Rapid bubble sorting and short sliding window are used in the new algorithm to sort the position of the code words according to their hit times, which can improve the probability of matching the active code word at the first time. And a short sliding time window is used to buffer the change of the pixel and trace the average and deviations, which can effectively solve the adaptive problem of the model in the case of dynamic background. The result of experiment shows that the improved algorithm provides better detection precision and real-time performance in complex environment conditions.