Smart mobility procedure for road traffic noise dynamic estimation by video analysis

Road Traffic Noise is a major concern in Europe, with more than 20% of people exposed to harmful noise levels. Efficient monitoring and assessment of the sound levels in critical areas are crucial to support decision strategies to control/reduce noise exposure. However, continuous and long-time ranged spatio-temporal measurements require high-cost equipment and maintenance duties. Therefore, this paper aims to develop a cost-efficient smart mobility procedure for the estimation of traffic noise levels based on roadside video images. The developed procedure involves an algorithm that extracts traffic volumes, identifies vehicle classes, estimates each vehicle's speed from video recordings, and a noise assessment component using dynamic microscopic models. These latter are based on existing Noise Emission Models – NEMs, for the assessment of the source sound power levels, coupled with a sound propagation model able to consider each on-road vehicle speed as input and evaluate the equivalent continuous A-weighted sound pressure levels. The developed approach is characterized by a modular structure that easily allows to replace NEMs and/or incorporate extra variables in the sound propagation model. The procedure is tested on a rural road of a medium-sized city, under different levels of service, and results show that the errors concerning the noise estimations are below 1 dBA, revealing high accuracy.