In this paper, we proposed a multi-module method for detecting human face from complex background. Each module utilizes one of the cues of human face, such as color, geometric property, motion, depth and etc.. Facial region is obtained by fusion of detected parameters result from each module, and the result is surely robust than that of using any single module. As an actual preliminary implementation, we constructed a face detection system using color module and geometric module. Although only two modules are realized in our system, the conceptual structure of the proposed multi-module method is through to be an efficient solution to the problem of face detection. Multi-module method can not only improve the robustness of detection, but also make the development of new applications, especially face detection systems an easy task, which only need to combine the extra modules with other pre-existing modules. Experimental results show that faces in complex background can be extracted efficiently with multi-module method.