Real Time Computer Vision

ABSTRACT:The SURVICE Engineering Company is proposing the development of an accelerated background/foreground separation process using massively parallel computing. The proposed approach combines a proven, mathematically robust method for background/foreground separation with massively parallel computing to realize a high-performance implementation that scales across computing devices, from workstations and desktops to tablet computers and smartphones. The proposed Phase I effort will culminate in a real-time system prototype that demonstrates the efficacy of high-performance computing as a basis for important computer vision tasks.BENEFIT:The proposed approach will accelerate background/foreground separation in video streams using a proven, mathematically robust technique that scales across a wide range of computing devices. The implementation will provide a scalable, high-performance background/foreground separation process that will serve as an integral step in detecting, identifying, tracking, and recognizing objects in video sequences.