An automated sign retroreflectivity condition evaluation methodology using mobile LIDAR and computer vision

Traffic sign retroreflectivity condition is one of the most critical factors impacting night-time driving safety. The federal rule on minimum traffic sign retroreflectivity incorporated in the 2009 Manual on Uniform Traffic Control Devices (MUTCD) issued by the Federal Highway Administration (FHWA) is compelling transportation agencies to evaluate how to comply. As the traditional manual methods have become financially and/or practically infeasible, there is an urgent need for an effective and efficient retroreflectivity evaluation method. This paper investigates the possibility and proposes a methodology for automatically evaluating traffic sign retroreflectivity condition using mobile light detection and ranging (LIDAR) and computer vision. The proposed methodology uses (a) the traffic sign detection and color segmentation methods that are introduced for the first time to evaluate the retroreflectivity of different traffic sign colors separately in an automated manner; (b) the proposed theoretical-empirical LIDAR retro-intensity normalization scheme to more reliably model the radiometric responses of traffic sign captured in the mobile LIDAR data; (c) the population-based condition assessment method for the first time to statistically quantify the retroreflectivity condition of traffic signs rather than an four-point average. An experimental test was conducted on 35 Type I Engineer Grade stop signs. The result shows that the proposed methodology can produce a promising outcome with consistent retroreflectivity evaluations based on handheld retroreflectometer measurements; the proposed methodology can, also, better identify traffic signs with non-homogeneous deteriorated retroreflectivity. The reliable retroreflectivity evaluation results make the proposed methodology an appealing alternative for transportation agencies to use to comply with the FHWA's requirement. (C) 2015 Elsevier Ltd. All rights reserved.