Effect of Environmental Conditions on Performance of Image Recognition-Based Lane Departure Warning System

Abstract

Lane departure warning systems (LDWS) provide an effective counter-measure against road departure crashes. Much commercial development of these systems has concentrated on image processing-based systems. Because these systems depend on the detection and tracking of lane markings, it is logical to hypothesize that the visibility of the markings under different weather and lighting conditions may significantly influence the performance of LDWS. Work was done to determine the effects of environmental conditions such as rain and lighting on LDWS performance with typical lane-marking installations on limited-access facilities. The main performance measure used in this study is the efficacy rate (ER). This is defined as the ratio of the number of instances in which the lane departure warning device is able to provide lane crossing alarms in all instances in which an LDWS-equipped vehicle crosses the lane markings. The results indicate that in most cases the ER of the LDWS under dry and light rain conditions is 100%. However, it was found that performance of LDWS of lane markings was affected significantly by heavy rain conditions at night for typical lane-marking installations. With heavy-rain night conditions, most of the observed ER values were between 0% and 30%. It appears that the ER associated with night rain conditions increases with increased lane-marking retroreflectivity. Dusk conditions were found to drop the ER of LDWS by 15% to 18%. At a few locations that had yellow lane markings near the end of their service lives or had yellow markings on concrete pavements, LDWS had difficulty detecting the markings because of the low level of contrast between the yellow markings and the pavement.