A Low-Cost IR-Enhanced Video System for Early Obstacle Warning

More than 6,000,000 auto accidents occur in the U.S. each year, with about one third resulting in injury and 40,000 resulting in fatality. Approximately thirteen percent of these crashes involve pedestrians. A much broader group of generalized roadway obstacles including pedestrians, other automobiles and non-fixed objects, is responsible for over ninety percent of the accidents. A reasonably priced, automated driver warning system for a wider range of obstacles could drastically reduce the number of accidents, injuries, and fatalities.Creare proposes an innovative combination of several technologies to provide early warning of roadway obstacles. It combines a CCD camera with sophisticated DSP-based image recognition t identify obstacles, while ignoring benign objects along turns and curves. Operation is similar to expensive military night vision systems, but relies on a judicious cost and performance optimization to provide a significant reduction in accident probability. By using a reasonably priced, infrared enhanced imager during darkness. In Phase I, we will demonstrate performance with off-the-shelf prototype hardware and showing real-time performance of the image recognition software algorithms to provide early warning.