An efficient approach of moving objects detection in complex background

Detecting moving objects is the first step of many video surveillance applications. Most existing simple background subtraction methods such as frame difference, running average (RA), and median filter, have the low computational cost, but they can't perform well in the complex scenes. Although some ordinary methods can do well in the complex scenes, they can't satisfy the real-time requirement because of its high computational cost. So in this paper, we propose an efficient approach for detecting moving objects, which has the low computational cost and high performance in the complex scenes. The proposed method first uses the running average algorithm and contour information to obtain moving regions roughly. Then an improved GMM algorithm is used to update the background model and detect foreground precisely. The experiment results show that our method has a lower computational cost and performs better both in the outdoor and indoor scenes than GMM.