An improved method of object fragmentation

Background modeling plays a key role of event detection in intelligent surveillance systems. Gaussian Mixture Model (GMM) is a popular background modeling method in latest surveillance systems. However, the model will result in object fragmentation if the objects´ color is likely to its background. In our paper, we present a different mechanism which compares the area of original pictures with background pictures in HSV color space. Experiments demonstrate that the proposed method can recover the lost part of objects caused by GMM model partly and enhance objects´ integrity comparing to GMM. A drawback of this method is that it may impact the speed of computing when there are too many moving objects in frames.