Article ID: | iaor20122368 |
Volume: | 22 |
Start Page Number: | 17 |
End Page Number: | 28 |
Publication Date: | Jun 2012 |
Journal: | Transportation Research Part C |
Authors: | Babari Raouf, Hautire Nicolas, Dumont ric, Paparoditis Nicolas, Misener James |
Keywords: | transportation: air, statistics: regression |
The measurement of atmospheric visibility is an important element for road and air transportation safety. We propose in this paper a novel estimator of the atmospheric visibility by already existing conventional highway cameras, with a technique based on the gradient magnitude selected by applying Lambert’s law with respect to changes in lighting conditions. The response of this estimator is calibrated by non‐linear regression with data from a visibility meter installed in a test site which has been instrumented with a camera. Through our technique, atmospheric visibility estimates are obtained with an average error of 30% for images taken in the day, with sky luminance between 10 and 8000cdm‐2 and visibility distances up to 15km. Our results allow us to envision practical implementation on roadsides in the near future to determine local visibility for the benefit of road safety, meteorological observation and air quality monitoring.