Along with the development of road traffic systems, video vehicle detectors that can gather various traffic conditions are much expected recently. However, conventional gray-scale based video detectors do not work well under adverse conditions such as shadows, rainy nights, twilight and congestion. To solve this problem, a stereo-based video vehicle detector, SilhouetteVision, was developed, and its superb vehicle detection accuracy has been confirmed by various field tests. The focus here is to evaluate the performances of SilhouetteVision under snow conditions. It is very difficult to preserve stable performances during periods of snowfall for conventional video detectors. Excellent performance of the stereo-based SilhouetteVision was obtained in snow. For the covering abstract see ITRD E114174.
Abstract