Skip to Main content Skip to Navigation

Détection des conditions de visibilité et estimation de la distance de visibilité par vision embarquée

Abstract : The adverse weather conditions, in particular the fog, pose a problem for the drivers, who tend to overestimate distances, but also to the exteroceptive sensors which work less well. The detection and the quantification of the visibility conditions thus constitute a strong stake in terms of road safety. We propose two complementary methods onboard a vehicle aiming to estimate the visibility distance by using real time image analyses techniques. Both are based on the definition of the meteorological visibility distance, which is the greatest distance at which a black object of suitable dimensions can be recognized by day against the horizon sky. The first technique, using a model of atmospheric diffusion, detects and estimates the density of daytime fog by using a single camera. The second technique, using a generic property of the atmosphere, is able to estimate day and night the visibility distance under all meteorological conditions by using a stereoscopic sensor. On one hand, we evaluate our methods, thanks to several video sequences grabbed under different meteorological conditions, what constitutes a qualitative evaluation of the work done. On the other hand, we evaluate the methods, thanks to actual images of a specific site equipped with reference targets, which has been conceived and realized on the test facilities of Satory, what constitutes a quantitative evaluation.
Complete list of metadatas

Cited literature [47 references]  Display  Hide  Download
Contributor : Nicolas Hautiere <>
Submitted on : Saturday, May 13, 2006 - 1:11:52 PM
Last modification on : Wednesday, November 20, 2019 - 2:55:38 AM
Long-term archiving on: : Monday, September 17, 2012 - 2:26:12 PM


  • HAL Id : tel-00068743, version 1



Nicolas Hautiere. Détection des conditions de visibilité et estimation de la distance de visibilité par vision embarquée. Traitement du signal et de l'image [eess.SP]. Université Jean Monnet - Saint-Etienne, 2005. Français. ⟨tel-00068743⟩



Record views


Files downloads