Real-Time Vision-Based Vehicle-to-Vehicle Distance Estimation on Two-Lane Single Carriageway Using a Low-Cost 2D Camera at Night

Abstract

Estimation of distance with great accuracy, especially at night where vehicles are not visible is really a challenging research area. This article presents a pleasant vision-based real-time assistance system on any two-lane single carriageway at night. In this paper, a complex traffic scenario is considered where both ongoing vehicles and oncoming vehicles run on a two-lane single carriageway. As the environmental condition is very dark at night, hence only headlights and taillights of vehicles are visible. The proposed assistance system can estimate the vehicle-to-vehicle distance of ongoing or oncoming vehicles with respect to the test vehicle at night. The distance of the slower vehicles in the front along with the oncoming vehicles are estimated by observing only the position of taillights and headlights respectively. Sensitive vision-based parameters of headlights and taillights are used to estimate the exact distance between vehicles. Numerous real-time experiments reveal that the estimation accomplishes greater accuracy over the state-of-the-art techniques using a low-cost 2D camera.