U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations
|This report is an archived publication and may contain dated technical, contact, and link information|
|Publication Number: FHWA-HRT-13-018 Date: April 2013|
Publication Number: FHWA-HRT-13-018
Date: April 2013
Daylight measurements of the color of retroreflective materials used for traffic control signs are generally more variable than daylight color measurements of uniform diffuse surfaces.(1) Retroreflective materials have spherical or prismatic elements that direct light back in non-uniform ways. These materials are primarily designed to reflect light from the headlights of a vehicle back toward the vehicle to improve nighttime visibility for the driver. The color of the material under daylight viewing conditions is usually of secondary concern.(1) The reproducibility of daylight color measurements of retroreflective materials in the field and in the laboratory shows considerable variability, with different measuring instruments yielding different results.(2) Further, instrument measurements of chromaticity often do not correspond to perceived color judgments made by human observers.(2) Understanding and reducing these inconsistencies are important to the Federal Highway Administration (FHWA) in defining the size and shape of the color areas used to specify colors for traffic control signs. These color areas, or color boxes, are incorporated into the Code of Federal Regulations (CFR) as color specifications for retroreflective sign and pavement marking materials.(3)
The color of an object depends on, among other parameters, the spectral properties of the illuminant. International Commission on Illumination (CIE) standard illuminant D65, which is representative of average daylight in the northern hemisphere, has a correlated color temperature (CCT) of 6,500 K. Illuminant D65 produces a substantially different spectrum than illumination from CIE standard illuminant A, which has a CCT of 2,856 K and is intended to represent typical tungsten-filament lighting. Illuminant A is traditionally used to make nighttime measurements of retroreflective materials as illuminated by the tungsten-halogen headlights of a vehicle. Therefore, in the United States, Federal regulations give color specifications of retroreflective sign materials that are different for nighttime and daytime conditions.(3)
For retroreflective materials, FHWA daylight chromaticity coordinates defining the acceptable color regions for the six colors used in the present experiment are plotted in the CIE 1931 x, y chromaticity diagram in figure 1.(4) The corresponding CIE chromaticity coordinates for the same six colors are shown in figure 2.(4) Important differences exist between the two sets of daytime color coordinates for retroreflective sheeting. The CIE coordinates specify a smaller color area for white and a somewhat smaller color area for orange and display a wider hue angle separation (more blank space between lines of constant hue) between the red, orange, and yellow areas. Although the relationship between color coordinates and hue discrimination by human observers is not precise, this wider separation is an indication of less potential color confusion in the CIE system. One of the purposes of the present experiment was to examine whether the smaller separation in the FHWA system is adequate to support color discrimination of retroreflective sign materials by human observers.
|Figure 1. Graph. FHWA daylight chromaticity coordinates for various colors of retroreflective materials plotted in the CIE 1931 color space.|
|Figure 2. Graph. CIE daylight chromaticity coordinates for various colors of retroreflective materials plotted in the CIE 1931 color space.|
The retroreflective properties of sign materials significantly improve the visibility of traffic control signs during nighttime viewing, but they also affect their color appearance and brightness during daylight viewing.(1) The present study uses four different retroreflective materials and a standard of diffuse reflection to investigate the effects of various retroreflective properties on the chromaticity and luminance of reflected light under natural daylight and simulated daylight (illuminant D65) conditions. The study compares these instrument measurements with the daytime color appearance of the materials as judged by a group of human observers. Although it is difficult to relate luminance to brightness in a precise manner, the study evaluates the degree of correlation existing between measured luminance and apparent brightness to gain insight into the relative brightness of these various retroreflective materials as they might appear to drivers under daylight conditions.
Abramov et al. developed a technique for specifying color appearance by human observers that has considerable advantages over traditional means such as tristimulus colorimetry.(5) The method does not require precise equipment or controlled viewing conditions and thus lends itself readily to field applications. The technique is based upon direct perceptual hue and apparent saturation scaling. Research participants give percentages of their sensations based on four unique hue names: red, yellow, green, and blue. They also give a separate achromatic percentage for apparent saturation. The results of these direct scaling determinations of color appearance are expressed on a uniform appearance diagram (UAD), an opponent color diagram composed of two orthogonal axes (red-green and yellow-blue). Abramov et al. related the results of their measurements to those of other color specification systems. In one example, the researchers were able to derive traditional measures of discriminability from UADs. In another example, wavelength discrimination could be predicted from the distances among stimuli on a UAD. In fact, the authors claim that the UAD method can yield discriminability data that are at least as metrically uniform as the 1976 CIE diagram.(5)
Gordon et al. expanded on this research and presented further evidence of the robustness of the technique.(6) The researchers provided detailed reasons for the choice of the four color names and claimed that participants do not need special training to use these terms, since the unique hue components refer to internal standards. They suggested using an arcsine transformation to reduce non-uniformities in variance due to having bounded scales between 0 and 100 percent. Gordon et al. explored the effects of long-term stability over several months, context provided by preceding stimuli, stimulus range, experience giving hue scaling judgments, and language. They found that the hue and apparent saturation scaling method was robust in the face of these effects and produced accurate and reliable data. They noted that about 5 percent of participants did not use the percentage scales appropriately and their data had to be excluded. The method described by Gordon et al. formed the basis for the color appearance ratings collected in the present study.(6)
Jacobs and Johnson conducted a study of the color appearance of different retroreflective pavement marking products.(7) Although the study was not directly concerned with retroreflective sign sheeting materials, the results may be relevant since pavement marking materials also involve both color and retroreflectivity. In their experiment, seven color-normal human observers made color rating judgments on a scale of 1 to 5 from white to yellow. The observers sat in a stationary motor vehicle and rated pavement marking samples at viewing distances from 39 to 118 ft (12 to 36 m). While the researchers did not use the method of hue and apparent saturation scaling, they did show a correlation of their rating scale with chromaticity measurements of different pavement marking products. The authors found significant differences in the color ratings for the various pavement marking materials. Daytime and nighttime color determinations were not the same. Some yellow pavement marking materials were rated as yellow under daylight illumination but white when illuminated with the low beam headlights of the test vehicle at night.(7)
Thomas-Meyers et al. presented findings indicating that a specific range of chromaticities acceptable for pavement marking colors could be useful for both daytime and nighttime viewing conditions.(8) The authors measured hue and apparent saturation ratings from 34 research participants who viewed colored pavement marking stripes, either yellow (center line) or white (edge line), on a background of pavement color presented on a computer display. Although there were minor differences, the color regions that were reliably judged as yellow or white were similar under both daytime and nighttime contrast conditions. Older and younger participants gave similar color appearance judgments, but color-deficient participants judged the colors somewhat differently, especially at night. Chromaticity coordinates for proposed Ohio standards for yellow and white pavement markings were consistent with color appearance data for the white color, but the yellow color needed to be shifted toward higher saturation values. When in-use pavement markings were compared to the proposed standards, the white markings were consistent, but some of the aged yellow pavement markings appeared to be white. Although Thomas-Meyers et al. employed a relatively large sample of 34 participants, they used less realistic viewing conditions. The stimuli were presented on a computer display, and the study concentrated on the color appearance of pavement markings rather than highway traffic signs.(8)
In 2007, Davis and Miller conducted a pilot field study for FHWA on the daytime color appearance of retroreflective materials used for highway traffic signs. The pilot field study led to the research described in the present report.(9) The researchers made physical measurements of the chromaticity coordinates and luminance for a small sample of FHWA color specifications (selected from those shown in figure 1) both in the laboratory and in the field. Five naïve observers used the hue and apparent saturation rating method described by Gordon et al. to rate the appearance of the colors.(6) Six different types of retroreflective materials were employed, along with a standard diffuse white reflector. The hue and apparent saturation results showed less saturation for both the yellow and orange retroreflective samples when compared with the diffuse reflector with the same colors. Based on limited data, the authors tentatively concluded that the instrument measurements were generally consistent with the color appearance judgments of human observers in terms of both hue and saturation.
The present study investigated the color appearance of retroreflective materials used for traffic control signs. The experiment compared field judgments of perceived hue, apparent saturation, and brightness made by human observers with instrument measurements of chromaticity, saturation (derived from the CIE 1976 L*a*b* (CIELAB) calculations), and luminance made with spectroradiometers in the laboratory and in the field.(10) While the term chroma is used to describe the concept of saturation in CIELAB, the term saturation is used interchangeably in this report. Perceptual and physical measurements were made for 120 color samples.
Although brightness is difficult to measure and to correlate with physical measurements made by instruments, brightness was included in the investigation because it could yield further insight into the perception of drivers regarding retroreflective sign material. This study expanded on the methodology of Gordon et al. by adding a brightness scale to the procedure.(6) The present study also included a larger participant sample (n = 17) than was used in earlier studies.(5, 6) In addition, the study employed samples of real highway traffic sign materials viewed under actual daylight field conditions, instead of small spots of light presented in a laboratory as in the originating studies. Gordon et al. reported somewhat low within-participant variability; however, participants in the earlier studies were experienced, having participated in several experiments, and rather homogenous, coming from an academic environment.6 It was uncertain how age and gender could affect variability for a sample of naïve participants recruited from the general driving public. The present study included a more heterogeneous group of participants, including both male and female as well as a wide age range.