VSAR (Viewers Situational and Spatial Awareness for Applied Risk and Reasoning) is a three-year project to develop tools that will combine live images with pre-generated 3D models. The project brings together some of the leading players in this emerging technology – NPL, BBC R&D, and BAE Systems – is led by C.A.S.T. and is co-funded by the Technology Strategy Board.
The focus of the project is on how best to take live information and images gathered from multiple cameras and overlay it on a 3D modelled environment. This environment might be anything from a rendering of the All England Tennis Club – which would incorporate live footage of the Wimbledon Championships – to a 3D model of a busy town centre overlaid with footage from an array of CCTV cameras.
In use, the technology provides a helicopter’s eyeview of the scene, allowing the viewer to fly around within an interactive mixed reality (of 3D model and real-time action). This greatly improves the viewer’s grasp of exact locations and the events taking place there.
As part of BBC R&D’s contribution to the project, we are extending our previous work on segmentation of objects from arbitrary backgrounds, and investigated new methods for merging live images into a 3D modelled environment. We’ve also developed real-time software for tracking the pan, tilt and zoom of a broadcast camera by analysing the movement of image detail (such as corners and edges), and computing its position in 3D space with respect to known features in the scene, building on our previous work in the MATRIS project.
~RS~a~RS~rd~RS~q~RS~~RS~z~RS~07~RS~){kind=small}
