Automated Mobile Highway Sign Retroreflectivity Measurement

This paper presents a mobile image-based system for measuring the brightness of highway signs at night time, so that visibility standards for signs can be evaluated and maintained to protect the safety of the motoring public. The system is composed of a high speed IEEE-1394 digital camera, and a laptop computer with a purpose-designed measuring software, which is composed of three modules: 1) Image and video processing; 2) Image analysis; and, 3) Data storage. Using vehicle headlamps as an illumination source, the system can acquire, classify, and analyze the visibility of the traffic signs in real time, and report results to a data file. The analysis algorithms, including elements of sign isolation, classification, and intensity measurement, are designed for easy implementation and fast real-time measurement. The results of static testing experiments on signs, and on-road real-time measurements are described in this paper and they show the feasibility of such a system. INTRODUCTION For highway signs to accompolish their intended purpose, which is to promote highway safety and efficiency, they must be visible to motorists. At night, the visibility of the signs depends on the retroreflectivity of the sheeting materials on the signs. This paper presents a mobile image-based measuring system for measuring the brightness of highway signs at night time, so that visibility standards for signs can be evaluated and maintained to protect the safety of the motoring public. Signs must be legible and color distinguishable at night as well as day. While this can be accomplished through external illumination of the signs, retroreflection is the most commonly used means of making signs visible to the driver at night (Black, et al, 1992). When illuminated by external lighting sources such as automotive headlamps, traffic signs appear bright in proportion to their ability to redirect the incident illumination back toward the driver. The term luminance is used to quantify the amount of light that is redirected by the sign (Paniati, et al, 1993). So luminance is “what the motorist actually sees”. Retroreflectivity is a term used to describe the amount of light reflected from retroreflective materials. It is measured in terms of the coefficient of retroreflection, Ra, which is the quotient of illuminance to the retroreflecting surface’s area (Paniati, et al, 1993). Ra is measured in units of candelas per foot candle per square foot (cd/fc/ft) or the metric equivalent, candelas per lux per square meter (cd/lux/m). Currently, measuring sign retroreflectivity in the field requires the use of a portable hand-held retroreflectometer, although sheeting material and new signs can be measured in a more accurate laboratory setting. While it is a portable unit, it is still too cumbersome and time consuming to be used for a wide scale inspection program. A mobile retroreflectometer is currently under development by the FHWA. It requires a laser range finder, a second operator, and electronic strobes, and results in significantly higher capital and maintenance costs. Road sign recognition (RSR) is a field of applied computer vision research concerned with the automatic classification of traffic signs in traffic scene images acquired from a moving vehicle (Paclik). A literature review has revealed that a significant number of papers deal with automated road sign recognition under daylight conditions, but very few papers deal with RSR at nighttime, only one paper proposes a mobile system. SYSTEM OVERVIEW The sign measurement system described herein was designed as a low cost system to be mounted in any vehicle, using nighttime images of signs illuminated by vehicle headlamps, thus making measurements that as best as possible reflect what the human eye can see. The system uses inexpensive, state of the art, off the shelf hardware to acquire the images. Purpose designed software acquires, classifies, and analyzes the brightness of the road signs in real time, and reports results to a data file.