Traffic Safety using Frame Extraction Through Time

In today's world the public understands video surveillance systems as a series of closed circuit television (CCTV) systems. People imagine tens of cameras connected to tens of remote monitors, controlled by multiple personal who pay attention to people, vehicles, and suspicious objects to improve overall public safety. Increasingly this view is incorrect as most systems are controlled and monitored using some form of computer vision system. Transportation safety is a key area where video surveillance is used for public safety. Canada's road safety vision is "to have the safest roads in the world". In order to provide this type of road safety a CCTV system called COMPASS has been deployed to monitor traffic. However, a computer vision monitoring system like this raises interesting and difficult problems for automated image processing. The changing lighting conditions have lead to algorithms which require a great deal of computational power to meet the needs of real-time operations and monitoring. In this paper we propose a new pre-processing approach to helping minimize nighttime lighting concerns by generating a nighttime view with daytime contrast from a sequence of image frames through time.