Automatic camera control in virtual environments augmented using multiple sparse videos

► Real-time rendering of an interactive augmented virtual environment for surveillance. ► Segmentation and tracking of moving objects using sparse cameras and the homographic constraint. ► Correction of color artifacts caused by cameras with significant color differences using simulated annealing. ► Generation of perspective corrected views of moving objects using pre-computed segmentation and tracking information, using projective texture mapping on billboards. ► Automatic virtual camera control to generate videos of selected subjects (3rd view) or display the scene viewed by a subject (1st person view), to facilitate monitoring and surveillance tasks.