Real-time moving maritime objects segmentation and tracking for video communication

Video object segmentation and tracking has been widely used in many video communication applications. This paper proposes an effective method of detecting and tracking the moving maritime objects in video sequences. An approximation algorithm, which adopts visual attention architecture to reduce the wavy noises and make the object distinct from the background, segments the regions of interest quickly. Then, a modified temporal differencing method is proposed to eliminate the background and detect the moving maritime object. On the multi-object tracking process, an adaptive choosing the global nearest neighbor algorithm is presented based on a linear dynamic model. Furthermore, an effective occlusion reasoning method is proposed, which partly solves the occlusion problem. Experimental results show that the proposed real-time system performs well in various maritime video sequences.