Single Gaussian model is an effective way for target detection under stable environment. However, it suffers from low robustness when there are dynamic scenes and/or sudden lighting changes. The correlation coefficients method is effective in describing the similarity between images. Furthermore, this method is not sensitive to small image appearance changes. To take advantage of this characteristic, a hierarchical block detection mechanism is proposed in this paper. First, the noise from the dynamic background scenes and/or the lighting changes are filtered by the correlation coefficients. After that, the blocks with foreground are segmented by single Gaussian model. Experiments confirmed that the proposed method is effective to deal with dynamic backgrounds and fast in computation.