A Spatiotemporal Motion-Vector Filter for Object Tracking on Compressed Video

In this paper, a novel filter for real-time object tracking from compressed domain is presented and evaluated. The filter significantly reduces the noisy motion vectors, that do not represent a real object movement, from Mpeg family compressed videos. The filter analyses the spatial (neighborhood) and temporal coherence of block motion vectors to determine if they are likely to represent true motion from the recorded scene. Qualitative and quantitative experiments are performed displaying that the proposed spatiotemporal filter (STF) outperforms the currently widely used vector median filter. The results obtained with the spatiotemporal filter make it suitable as a first step of any system that aims to detect and track objects from compressed video using its motion vectors.