A method for background modeling and moving object detection in video surveillance

In this paper, we proposed an algorithm based on sigma-delta filter that gives good performance on background estimation and foreground detection. With efficient operations, delta-sigma filter can track the changes of background. There are some recent studies been carried out, however, they focus on intensity variance estimation, and in their works estimation and detection are highly coupled. In our work, global variance is used to update background pixels. The influence between background estimation and foreground detection is reduced. Background model and intensity variance are updated selectively and partially to make a good balance between sensitivity and reliability. Since the initialization sensitivity, a preprocessing step is introduced. Proposed algorithm is tested by detecting moving objects in street and subway sequences. In the test, our algorithm works efficiently, and result demonstrates the effectiveness and benefits of the proposed algorithm.