High-Precision Dynamic Traffic Noise Mapping Based on Road Surveillance Video

Abstract

High-precision dynamic traffic noise maps can describe the spatial and temporal distributions of noise and are necessary for actual noise prevention. Existing monitoring point-based methods suffer from limited spatial adaptability, and prediction model-based methods are limited by the requirements for traffic and environmental parameter specifications. Road surveillance video data are effective for computing and analyzing dynamic traffic-related factors, such as traffic flow, vehicle speed and vehicle type, and environmental factors, such as road material, weather and vegetation. Here, we propose a road surveillance video-based method for high-precision dynamic traffic noise mapping. First, it identifies dynamic traffic elements and environmental elements from videos. Then, elements are converted from image coordinates to geographic coordinates by video calibration. Finally, we formalize a dynamic noise mapping model at the lane level. In an actual case analysis, the average error is 1.53 dBA. As surveillance video already has a high coverage rate in most cities, this method can be deployed to entire cities if needed.