Evaluation of Video Detection Systems, Volume 4: Effects of Adverse Weather Conditions in the Performance of Video Detection Systems

摘要

The performance of three video detection systems (VDS): Iteris, Autoscope, and Peek, was evaluated using a side-by-side installation at a signalized intersection under various adverse weather conditions including rain and snow in both day and nighttime, and light fog and dense fog in daytime. Four types of detection errors (false, missed, stuck-on, and dropped calls) were quantified at stop bar and advance detection zones, and all errors were visually verified using video images from the intersection. Results indicate that VDS performance was not greatly impacted under daytime light fog or rain conditions without wind, but significant changes were observed under dense fog and snow in daytime, and snow and rain in nighttime. During dense fog conditions Iteris and Autoscope changed their operating mode and placed constant calls due to image contrast loss, with potential for significant effects on traffic operation, while Peek significantly increased its missed calls. Snow in daytime and nighttime greatly increased false calls for the three systems (more than 50% of the VDS calls were false), but it had limited effects on missed, stuck-on, and dropped calls. False calls during rainy conditions without wind also increased, during both daytime (around 10%) and nighttime (between 10% and 50%), mostly due to the reflection of headlights of vehicles in the adjacent lanes. Detailed performance of the three VDS in each of the three stop bars and the three advance zones, as well as potential causes and effects of the most prominent error types are presented.