 |
| Robust Movement Detection in Outdoor Scenes | |||||
| Participants Yi Liu M.Syst.Sc. student 2001-2003 Dr. Pierre Payeur SITE, University of Ottawa Collaborators  Natural Sciences and Engineering Research Council of Canada  Canadian Foundation for Innovation  Ontario Innovation Trust |
Nowadays video surveillance systems are introduced in our world at a growing pace. Several systems have been proposed to monitor
activity of people. Applications to traffic monitoring have also been developed. But while computer vision techniques
are fairly reliable when they operate in a controlled environment, inside a building for example, several difficulties are still
faced when cameras are installed in outdoor locations. This is especially true in nordic climate where lighting and weather conditions
are far from being uniform throughout the year. Abrupt variations in lighting due to time-of-the-day and clouds movement, shadow
effects, as well as changes in visibility resulting from fog, rain, snowfalls, or pollution, severely impede vision systems capability
to detect and track activity. This project involved the development of image-based movement detection systems that are robust to harsh outdoor conditions. Strategies have been elaborated and tested to monitor color map information in such a way that the vision system dynamically switches between various detection modes, according to the prevalent lighting and weather conditions. The implementation has been built around a traffic monitoring system meant to anticipate vehicles arrival at an intersection in order to optimize traffic lights control. Depending on the detection mode in operation, various features are monitored and shadow effects are compensated when potentially impeding the behavior of the system. Tests conducted on real video sequences of outdoor traffic scenes in Ottawa, Canada, demonstrated the robustness of the proposed framework under various weather conditions and time-of-the-day operation.
|
Related Publications
|
© SMART Research Group, 2008