Road Width Estimator—An Automatic Tool for Calculating Road Width Leveraging Geospatial Knowledge

The emergence of geospatial knowledge reinforced with computer vision to extract meaningful insights from remotely sensed images has achieved remarkable success. Transport infrastructure, being one of the promising fields to showcase various implementations of these technologies, remains unexplored in numerous aspects. This study presents a road width estimation tool leveraging the geospatial information embedded into remotely sensed imagery. The proposed Road Width Estimator (RWE) tool is developed in Python, utilizing Canny Edge detection and Douglas-Peucker algorithms. This tool can effectively compute width at different junctions and road segments for various road types. Extensive experiments on satellite and aerial images demonstrate its ability to accurately estimate road widths for different built-in scenarios and environmental conditions. The findings show RMSE to be less than 0.1 m and a correlation coefficient value of 0.97, demonstrating a high positive correlation between derived widths and onsite estimated width. The validation results showcased that the RWE tool could be adequately applied to dense infrastructure or rural settings with limited resources, offering valuable solutions for road width estimation challenges.