Color recognition of retroreflective traffic signs under various lighting conditions.
Sign colors contribute substantially to correct sign recognition. While daytime color appearance is influenced somewhat by weather conditions (e.g., sun, clouds, rain, or fog) nighttime sign colors are greatly affected by the artificial lighting (headlamps, fixed sign lights, or a combination thereof) under which they are seen. For example, signs made from retroreflective materials reflect light from headlamps back to the driver very efficiently. Signs made from other than retroreflective materials (i.e., from opaque materials) are not adequately visible under headlights alone and must be illuminated by fixed lights to be visible at night. Fixed sign lights are also often used with retroreflective signs to improve the signs' conspicuity and legibility. In such cases signs are seen under a combination of fixed sign lighting and headlights. Depending on the spectral distribution of the light produced by the various types of sign lights and headlamps, the chromaticity of illuminated signs will shift markedly at night. The present research grew out of concern over possible changes in color appearance resulting from such chromaticity shifts as well as from changes in color specification. The study was conducted in 1990 at the Turner-Fairbank Highway Research Center and was sponsored by the Federal Highway Administration (FHWA) as part of the Grants for Research Fellowships program.
Color Specifications for Traffic Signs
Color coding is used to aid drivers in the prompt recognition of highway signs, pavement markings, traffic signals, and other traffic control devices. The Manual on Uniform Traffic Control Devices (MUTCD) specifies meanings for eight colors:
* Red - stop or prohibition. * Yellow - general warning. * Orange - construction and maintenance warning. * Blue - motorist services guidance. * Green - indicated movements permitted, direction guidance. * Brown - recreational and cultural interest guidance. * Black and white - regulatory.(1)
To standardize the appearance of the colors used in the MUTCD, the FHWA has issued color specifications in the form of central values and tolerance limits. Historically, these specifications were based on the colors of available materials rather than on experimental data on color recognition. For example, the acceptable region for yellow includes the rather orangish yellows produced by lead chromate, the pigment once widely used in producing yellow road markings. Progress in the manufacture of pigments and retroreflective materials may have made an improved standard highway yellow possible.
There has been concern about possible color confusion among red, orange, and yellow sign colors, especially at nighttime, since the FHWA specifications for these three colors are relatively closely spaced in the International Commission on Illumination (Commission Internationale de l'Eclairage-CIE) chromaticity diagram. Several recent studies have analyzed experimental data on the color appearance of materials viewed under diffuse sources of illumination.[2,3,4] These materials included ordinary (opaque) paints and pigments conforming to the American National Standards Institute's (ANSI) specifications Z-53.1 as well as retroreflective samples conforming to FHWA specifications. One analysis recommended that the safety color specifications should be considered for adoption in highway signing applications.
Currently used tungsten-halogen headlamps produce light with much energy in the red portion of the spectrum with a related color temperature of around 2,800 to 3,000 K (figure 1). Daylight, on the other hand, contains much energy in the blue portion of the spectrum - about 6,500 K for natural daylight (figure 2). Since signs (or any other object for that matter) can only reflect light available, their colors appear different under headlights than under daylight. Motorists are, however, accustomed to this much redder appearance. Within the next 2 years, cars will probably be produced with metal-halide headlamps rather than tungsten-halogen headlamps. The industry's rationale for changing from tungsten-halogen to metal-halide is the latter lamp's much higher efficiency, longer life, and smaller size (figures 3 and 4.) The metal-halide lamp produces visible energy at distinct wavelengths across the spectrum with a good amount of blue and considerably less red than the tungsten-halogen lamp; its related color temperature is expected to be about 4,200 to 4,500 K (figure 5). The effect of such a change in headlamp spectral composition on the nighttime appearance of traffic signs is not yet fully known. One computational analysis found mixed results for the sample of High-Intensity Discharge (HID) spectral power distributions studied; there were indications that the discriminability of red versus orange, yellow, and brown might actually be improved under certain HID sources as compared with tungsten halogen sources.
The main objective of the present study was to compare current FHWA standard retroreflective signing materials with those materials following ANSI safety color specifications. Specifically, the study: * Compared color identification of these materials when viewed in the retroreflective mode (i.e., under headlighting) as well as under diffuse lighting.
* Compared color recognition for retroreflective materials viewed under tungsten-halogen versus metal-halide simulated headlamps, either alone or in combination with various fixed light sources commonly used for sign lighting.
The laboratory setup was designed to simulate sign-viewing geometry for a passenger car positioned 400 ft (121.9 m) from a traffic sign, with the sign 14 ft (4.3 m) to the right from the right-hand edge of the road and 8 ft (2.4 m) above the road surface (figure 6). For that geometry, the driver sees the sign at an observation angle of about 0.2 degrees. The observation angle is comprised of a ray from the headlights to the sign and a ray from the sign back to the observers' eyes.
Twelve different retroreflective sheeting materials (7 yellow, 3 red, and 2 orange) from 2 manufacturers were evaluated. Since two pilot studies had shown no problems in color recognition of blue and green samples, this study was restricted to yellow, red, and orange.[7,8] The FHWA and experimental ANSI samples were included for each color.
Since retroreflective sheeting is not commercially available in ANSI colors, these samples were custom made for use in this study. The red and orange sheetings were all FP-85 Type II (enclosed lens) materials; the yellow sheetings included Type III (encapsulated lens) and an experimental prismatic material (not included in the FP-85 typology), as well as Type II materials. See table 1 for a full description of the 12 sheeting materials. Each of the 12 materials was mounted on a 1- by 2-ft (30- by 61-cm) aluminum panel. Black and white striping was used as a border along the upper and lower edges of the panels; otherwise, the panels were blank.
Three types of lighting were used: simulated daylight, headlighting only, and a combination of headlighting plus fixed sign lighting. A daylight simulator with two spectrally correct fluorescent lamps of 6,500 K was used to provide the diffuse simulated daylight condition.
The light sources simulating either tungsten-halogen or metal-halide headlamps had to be quite small to fit into a space that allowed maintenance of the 0.2-degree observation angle within the 50-ft (15.2-m) laboratory space. This size limitation precluded the use of real headlamps.
To simulate the tungsten-halogen headlamp, a small 30-W tungsten-halogen narrow beam floodlamp was used. To simulate the metal-halide headlamp, a 70-W neutral white metal-halide lamp with a 3,800 K color temperature was used. Figure 5 shows the spectral power distribution for this lamp. The lamps were located close to the observers' eyes (one to the left, the other to the right) to ensure maintenance of the 0.2-degree observation angle. Photometric information on sealed beam headlights showed an illuminance level of 0.2 lux at the sign surface. To provide this illuminance level at the sign surface, neutral density filters were used to attenuate the luminous intensity of these lamps.
For the fixed sign lighting used in combination with headlighting, four different light sources were used: clear mercury, phosphor-coated mercury, phosphor-coated metal-halide, and high-pressure sodium. Each of these lamp types was mechanically adjusted to produce about 200 lux on the sign surface with a geometry similar to that used for bottom-mounted sign lights. This level represents the average illuminance used for overhead guide signs.
Each of the 12 signing materials was viewed under a total of 11 lighting conditions:
* One daylight condition. * Two headlight type conditions. * Eight combination conditions: two headlight type conditions with each of four different fixed sign lights.
Forty subjects participated in the study. They were divided among four different age groups - 25 and under, 26 through 44, 45 through 65, and 66 and over. Each age group contained five males and five females. All subjects were licensed drivers with corrected visual acuity of 20/40 or better, as tested with a Bausch & Lomb Ortho-Rater. Subjects were screened for normal color vision with the Ishihara pseudoisochromatic plate test.
Subjects were seated on a chair of adjustable height and given time to adapt to the lighting conditions. Subjects were instructed to press a handheld button as soon as they could identify sign color; they were told to use a single-word color term such as green, yellow, red, lime, or olive in making their identification. To begin each trial, a single sign panel was moved into the viewing position; the shutter was then opened. The subject's color naming response and response time were recorded. A within-subjects design was used in which each subject viewed each of the 132 stimuli (i.e., 12 panels under each of the 11 lighting conditions).
Data Analyses and Results
For purposes of analysis, color naming responses were coded as correct if they agreed with the nominal color of the material; any other response was considered incorrect even though there is no real right or wrong regarding a person's perceptions. Thus, if a subject identified a nominally orange sample as gold, this was coded as an incorrect response.
Table 1 presents the percentage of subjects responding correctly to each color sample under each of the 11 lighting conditions. Figures 7 through 9 are bar graphs based on these data. Figure 7 shows overall means for FHWA and ANSI samples for each color, averaged across all 11 lighting conditions. In figure 8, the means are averaged across the daylight and headlights only conditions. Figure 9 presents the means across the eight headlights with fixed lighting conditions. [Tabular Data Omitted]
Two general features of the data may be noted. First, consider the three general types of lighting used. As would be expected, simulated daylight resulted in more accurate color naming than did either the headlights only or the fixed lighting with headlights conditions. Second, consider the relative accuracy of recognition for the three colors studied. The red samples were the best recognized of the three colors. (Under daylight, 100 percent of the subjects' responses for the reds were correct.) In most cases, the orange samples were also correctly identified by a large majority of the subjects. The yellow samples were, on the whole, correctly identified less frequently than the reds or oranges, regardless of the lighting condition.
The study's most important finding was that while there was variation within the material-lighting combinations, the type of specification (FHWA or ANSI) did not result in any large or systematic difference in the level of accurate color identification. This finding held true regardless of the lighting condition considered.
The second major finding was that the accuracy of color identification was not systematically affected by the type of headlamp used. This finding can be seen by comparing the mean percentages (averaged across all 12 materials) for tungsten-halogen (63.8 percent) and metal-halide (63.8 percent), as well as by comparing the individual percentages for each material under these two lighting conditions.
Discussion and Conclusions
This study aimed to assess the possible improvement in driver recognition of traffic sign colors to be achieved by changing from the FHWA highway colors to the ANSI safety colors. To this end, sign color appearance under daylight and various nighttime viewing conditions was investigated. Although some lighting-color combinations improved correct color recognition when viewing the ANSI safety colors, these improvements were marginal and applied only to some of the signing material colors and types.
Several caveats must be recognized in discussing study findings. For example, one uncertainty exists regarding spectral correctness of the ANSI experimental samples, since these were hand-made in a laboratory shop. There are no color specifications for retroreflective ANSI safety colors at this time: the ANSI Z-53.1 standards to which the samples were made apply to opaque (nonretroreflective) paints.
Another uncertain variable was the metal-halide lamp used in the experiment. The Society of Automotive Engineers' Discharge Forward Lighting task force and CIE's Technical Committee "Headlighting with HID Lamps" are still working to develop a standard for HID headlamps that includes, among many other parameters, color specifications for the metal-halide automotive headlamp. This study used a commercially available small wattage metal-halide lamp similar in makeup to what the final headlamp design is expected to be. Although study findings thus most likely resemble what would be found using the final headlamp design, some caution should be exercised in using the results of this limited study to recommend color specification changes.
Finally, the appropriateness of the various lighting configurations used during data collection must be addressed. Red, yellow, and orange signs are predominantly seen under either daylight or headlamp illumination only: with the exception of some construction signs, signs of these colors are seldom seen under fixed sign lighting. Thus, correct identification of the red, yellow, and orange FHWA colors versus the ANSI colors is most significant under daylight or headlight only conditions. While the present nighttime study did not show any substantial benefits of the ANSI safety colors over the currently used FHWA highway colors, other studies did show significant benefits when using the ANSI colors under diffuse daylight conditions. Therefore, additional work should be conducted once a prototype of the finalized metal-halide headlamp design and appropriate retroreflective ANSI signing material samples are available. This work should investigate not only the color recognition of these materials, but also their overall effectiveness relative to road safety. (1) Italic numbers in parentheses identify references on page 7.
 Manual on Uniform Traffic Control Devices (MUTCD). U.S. Department of Transportation, Federal Highway Administration, Washington, DC, 1988.
 B.L. Collins, B.Y. Kuo, S.E. Mayerson, J.A. Worthey, and G.L. Howett, Safety Color Appearances Under Selected Light Sources, National Bureau of Standards Internal Report 86-3493, Gaithersburg, MD, December 1986.
 B.L. Collins. Evaluation of Colors for Use on Traffic Control Devices, National Institute of Standards and Technology Internal Report 88-3894, Gaithersburg, MD, November 1988.
 S.F. Hussain, J.B. Arens, and P.S. Parsonson, (1989). Effects of Light Sources On Highway Sign Color Recognition. Transportation Research Record 1213, Transportation Research Board, National Research Council, Washington, DC, 1989.
 American National Standards Institute (ANSI). Safety Color Code for Marking Physical Hazards. New York, ANSI Z53.1, 1979.
 C.J. Simmons, M. Sivak, and M. Flannagan. Colors of Retroreflective Traffic Sign Materials When Illuminated by High-Intensity Discharge Headlights, Report No. UMTRI-89-7, University of Michigan Transportation Research Institute, Ann Arbor, MI, March 1989.
 R. Saremi. "Determination of Human Visual Capabilities in the Identification of the Color of Highway Signs Under a Combination of Vehicle Headlamp and High Intensity Discharge Light Sources." Dissertation, Oregon State University, Corvallis, OR, 1991.
 C.J. Simmons, Color Recognition for Blue, Green, and Yellow Retroreflective Materials Under Various Light Sources. Internal Report, Federal Highway Administration, Washington, DC, 1991.
 Standard Specifications for Construction of Roads and Bridges on Federal Highway Projects, FP-85, U.S. Department of Transportation, Federal Highway Administration, Washington, DC, 1985.
 S. Ishihara. Isihara's Test for Colour Blindness. Kanehara & Co., Ltd, Tokyo, 1989.
PHOTO : Figure 1. - Spectral power distribution for incandescent tungsten-halogen lamp used in this study.
PHOTO : Figure 2. - Spectral power distribution for typical daylight (CIE 1971).
PHOTO : Figure 3. - Arc-tube of proposed metal-halide automotive headlamp compared with a conventional tungsten-halogen lamp.
PHOTO : Figure 4. - Headlamp design utilizing metal-halide discharge lamp.
PHOTO : Figure 5. - Spectral power distribution for metal-halide lamp used in this study.
PHOTO : Figure 6. - Experimental design setup.
PHOTO : Figure 7. - Mean correct color identification across all lighting conditions.
PHOTO : Figure 8. - Mean correct color identification under daylight and headlights only.
PHOTO : Figure 9. - Mean correct color identification under fixed lighting with headlights.
John B. Arens is an illumination engineer in the Information and Behavioral Systems Division, Office of Safety and Traffic Operations Research and Development (R&D), Federal Highway Administration (FHWA). He is involved with research dealing with nighttime visibility and roadway safety. Before joining the FHWA, Mr. Arens spent more than 20 years with the Outdoor Lighting Division of Westinghouse Electric in various engineering and marketing positions.
A. Reza Saremi is an engineering systems analyst working for AEPCO, Inc., at the Turner-Fairbank Highway Research Center (TFHRC) in McLean, Virginia. Dr. Saremi's work involves data analysis as well as statistical analysis for the Highway Safety Information System. Prior to his present position, Dr. Saremi served as a graduate fellow at the TFHRC where he conducted this study.
Carole J. Simmons is an engineering research psychologist in the Information and Behavioral Systems Division, Office of Safety and Traffic Operations R&D, FHWA since 1990. Prior to this, she conducted human factors research, including a colorimetric analysis of retroreflective sheeting, at the University of Michigan Transportation Research Institute. She holds a doctoral degree from the University of Michigan.
|Printer friendly Cite/link Email Feedback|
|Author:||Arens, John B.; Saremi, A. Reza; Simmons, Carole J.|
|Date:||Jun 1, 1991|
|Next Article:||How to conduct questionnaire surveys.|