The purpose of the study is to construct a comprehensive background in complex scenes. This comprehensive system not only can utilize background subtraction to detect moving objects, the background of scenic and street view can also be constructed. The background reconstruction of the research can be divided into three steps. First, alternate background can be obtained by classifying groups which are selected using random image subtraction. Next, use the background updating technique to increase alternate background accuracy. Finally, focus on the dynamic background and process it in order to construct the correct image background. The experiment results showed the research can construct a complete background in complex scenes.