Real-time Detection of Slow Moving and Temporarily Stationary Objects

Detection of moving objects is an important research topic in visual tracking applications. Among the moving object detection algorithms, the background subtraction based method is widely adopted due to advantages such as less computational load and high detection quality. However, it is sensitive to illumination changes and fluctuating backgrounds. Moreover, it cannot deal with slow moving and temporarily stationary objects. Although numerous works concerning background model initialization and adaptation have reportedly overcome the obstacles associated with illumination changes and fluctuating backgrounds, however detection of slow moving and temporarily stationary objects remains a challenging issue. Hence, this paper is aimed at developing a novel real-time motion detection algorithm that can deal with the aforementioned problem. Experimental results demonstrate the effectiveness of the proposed approach.