Video analytics for panning camera in dynamic surveillance environment

Video analytics is key to intelligent surveillance systems ranging from simple algorithms of target detection to complex methodology for video semantics analysis. This paper presents a novel video analytics approach to generate a full view mosaic for a surveillance sequence to help detect moving objects in the presence of camera panning motion. Unlike the current panoramic transformation methods, in which pure horizontal panning motion is required to construct the mosaic, our method can deal with uncontrolled panning motion with slight camera tilt motion. This method provides a generic solution for real world surveillance camera issues such as undesired tilt motion because of the nonlinearity of camera moving mechanism, strong wind etc. Also this method can deal with both global and object motion. The sprite generated helps to detect moving objects, constructing motion trajectories and also provides a clean background reference even for a panning camera. The algorithm has been evaluated for several test sequences.