Virtu4D : A dynamic audio-video virtual representation for surveillance systems

The new generation of multimodal surveillance systems concurs to provide an automatic analysis of controlled environments, and a real-time interpretation of the scene and the need of situation awareness became more and more essential for systems dealing with surveillance of critical infrastructures to prevent potentially catastrophic incidents. When considering two heterogeneous information processes, one key issue is how to represent efficiently these two information in a unique representation. The following work tries to answer the issue of data representation by building a common perceptual space for heterogeneous surveillance systems and providing a common representation of the monitored environment (relying on acquired images and sounds). The main contribution of the paper is to propose a common perceptual space gathering audio and video information able to generate a real-time unique perception of the scene. Virtu4D, provides a flexible interface with a natural visual information display by making a virtual copy allowing a free intuitive 3D navigation in the environment and hence making the supervision system more intuitive and natural. This paper presents a short overview of the system along with the different algorithms involved in each step of the process.