Answers are grounded in this paper and cite their sources.
Efficient and scalable assessment of roadway lighting is important for supporting nighttime infrastructure inspection and maintenance. Conventional point-based photometric measurements can be time-consuming and provide limited route coverage. This study develops a smartphone-based video processing workflow for route-based screening of roadway lighting conditions using a smartphone mounted on a moving vehicle. The workflow combines ISO 14524-based camera calibration, continuous vehicle-based video acquisition, five-region image-plane segmentation, synchronised GPS positioning, and geographic information system-based visualisation. Greyscale intensities extracted from roadway image regions were converted to condition-specific luminance and illuminance estimates using a calibrated opto-electronic conversion function. The paired observations obtained from compressed video on an in-service asphalt surface over a reference illuminance range of 3.4 lx to 13.6 lx produced a root mean square error of 0.53 lx, a mean absolute percentage error of 6.2%, and 95% limits of agreement of approximately −1.00 lx to +1.17 lx under the tested configuration. Field application at two highway interchanges demonstrated the ability of the workflow to identify repeated route-level brighter and darker patterns and relative image-plane lateral variation. Field-derived means differed descriptively from the corresponding commissioned nominal full-output design averages by approximately 39% to 44%; however, the operating and adaptive-dimming status of the luminaires was not independently verified. Accordingly, the workflow offers a rapid and scalable route-level screening approach that combines broad route coverage with demonstrated quantitative agreement under the tested configuration, complementing standardised photometric measurements by efficiently identifying and prioritising locations warranting targeted investigation.
No takes yet. Share an insight, caveat, or question.
Golkarfard et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: