An effective detection algorithm for moving object with complex background

The temporal difference method can fast extract moving objects, but easily causes small holes and is generally not an effective method for extracting the entire shape of the moving object as well; the background subtraction method can extract the regions covering the moving objects preferably but easily to be affected by stationary objects in the scene that start to move. What's more, the adaptive subtraction of the background is very difficult to achieve. This paper proposes a new detection method for moving object with complex background. First of all, the filtering is done in both the temporal field and the spatial field to reduce the influence of noise. Secondly, the probable moving regions are fast subtracted by the temporal difference method. Finally, the object is precisely fixed by using Gaussian distributions of the adaptive mixture model (GMM). The experiments show that the method given in this paper is more efficient in extracting the moving object compared with temporal difference and GMM.