Skip to contentUnited States Department of Transportation - Federal Highway Administration FHWA Home
Research Home
REPORT
This report is an archived publication and may contain dated technical, contact, and link information
Publication Number: FHWA-HRT-13-044
Date: August 2013

 

Traffic Control Device Conspicuity

Sign Detection Conspicuity

Recently, Wertheim proposed a standardized and repeatable method of measuring conspicuity.(7) His method is based on the principle that conspicuity is not a property of the object but, rather, a property of the object and its environment. The measure combines the concepts of lateral masking and conspicuity, and it takes into account effects of environmental clutter on conspicuity. Wertheim's method, which expands on the work of Kooi and Toet, measures the extent to which individuals can divert their gaze from an object (i.e., the visual angle) and still detect the object or a property of the object.(46) For example, individuals might be asked to gradually move their fixation point away from a traffic signal until they can no longer detect its presence (detection method) or no longer identify the color of the signal indication (identification method).

Although Wertheim describes this method as standardized and repeatable and even suggests it could be used by agencies to regulate conspicuity, it is a psychophysical method that requires human observers as a critical instrument in the measurement procedure. Thus, the measurements are subject to individual differences in visual capabilities and may be subject to response bias. In addition, Wertheim describes three different implementations of the method that use different devices to measure the gaze diversion angle. While the results of each of these implementations are correlated, each yields significantly different conspicuity angles. Although the various methods yield different angles for the same stimulus, they all appear to yield results that are at least ordinal when ranking the conspicuity of different stimuli.

Wertheim has only applied his approach in a field setting, where conspicuity is assessed in the actual environments of the stimuli. Testing of TCD conspicuity in a laboratory environment is desirable because of the challenges involved in testing traffic signs in their natural environment. To use Wertheim's method to evaluate the conspicuity of a warning sign, observers would need to be positioned in a roadway traffic lane and gradually move their gaze away from the sign until it could not be detected. This would be a difficult and dangerous task in the presence of traffic. Therefore, a variation of the approach was developed that might be applied to TCDs in a laboratory setting and was validated by field observations that closely followed one of Wertheim's methods.

In the field, Wertheim's approach was implemented by asking participants to gradually shift their gaze away from actual signs in a daylight environment until they could no longer detect the presence of the signs. The laboratory approach used a staircase modification of the method of limits to approximate Wertheim's method with photographs of the TCD stimuli and environment.

In the modified laboratory application of Wertheim's methodology, panoramas 1 and 3 from the MDS classification were projected onto a screen for detection angle measurements (see figure 1 and figure 3). In addition, two panoramas of the environments used for the outdoor daylight assessments were used in the laboratory to enable direct comparison of the alternative approaches.

Method

Three signs served as detection targets in both the laboratory and outdoor environments: a speed limit sign (MUTCD, R2-1), a yellow pedestrian crossing warning sign (W11-2), and a fluorescent yellow-green pedestrian warning sign (W11-2).(36)

In the laboratory, the signs were presented against four backgrounds: an urban roadway scene (figure 1), a suburban background (figure 3), a copse (figure 23 and figure 24), and a parking lot (figure 25 and figure 26). In the outdoor environment, the actual copse and parking lot used in the laboratory projections served as backdrops for detection of physical signs.

A panoramic photograph is shown. An area with a green lawn in the foreground is horizontally bisected by a walkway. Beyond the walkway is a stand of trees. The trees have the appearance of early spring, some with flowers and others with pale green leaves. A two-story building with a flagpole in front is on the far left.
Figure 23 . Photo. Copse background without speed limit sign.

A panoramic photograph is shown. The panorama shows the same area as figure 23 except a 35 mi/h speed limit sign is shown in front of the stand of trees.
Figure 24 . Photo. Copse background with speed limit sign.

A panoramic photograph is shown. An area with a green lawn in the foreground is vertically bisected by a walkway. A parking lot full of cars can be seen beyond the lawn and walkway.
Figure 25 . Photo. Parking lot background without speed limit sign.

A panoramic photograph is shown. The panorama shows the same area as figure 25, except a 35 mi/h speed limit sign can be seen with the parked cars directly behind it.
Figure 26 . Photo. Parking lot background with speed limit sign.

Participants

The same 13 individuals participated in the outdoor and laboratory measurements. There were seven female participants (mean age 31 years, range 19-56) and six males participants (mean age 47 years, range 30-67). All participants were licensed drivers with corrected foveal visual-acuity in each eye of 20/30 or better. Seven participants completed the laboratory task first and six completed the outdoor task first.

Outdoor Procedure

A 36-inch (0.9-m)-tall speed limit sign and two 36-inch (0.9-m) pedestrian warning signs, one yellow and one fluorescent yellow-green, served as stimuli. Participants stood 85 ft (26 m) from a sign that was placed as shown in figure 24 and figure 26. Participants were instructed to point to the sign and then slowly point away, leftward, while gazing where they were pointing. They were to continue to move their gaze and point until they could no longer detect the sign in their peripheral vision. At that point, participants were asked to remember the location they were pointing when the sign was no longer detectable and rotate a compass pointer until it was aligned with that location. A researcher recorded the compass deflection angle and returned the pointer to zero. The procedure was then repeated two additional times with the same sign. This process was repeated with the other two signs for a total of nine trials.

Each participant completed the nine-trial procedure twice, once with the copse in the background and once with the parking lot in the background. The order of testing of signs and backgrounds was fully counterbalanced across 12 participants. However, one additional participant was tested and because no order effects were evident, the data for all 13 participants were included in the analyses.

Outdoor Stimuli

Photometric measurements were taken of the outdoor stimuli. The photometric measurements were made only once. Thus, because of varying times of participant testing and varying cloud cover, the measurements presented are intended only as estimates of the lighting conditions. These measurements also enable a rough comparison between the outdoor and laboratory lighting conditions. The laboratory measurements are reported in the next section. All measurements were taken from the location where participants stood when making conspicuity judgments.

Luminance measurements were made with a Konica Minolta CS-2000 Spectroradiometer. The white portion of the speed limit sign measured 555 fl (1,900 cd/m2) and 849 fl (2,910 cd/m2) with the copse and parking lot backgrounds, respectively. Average luminance of the areas to the right and left of the signs (0.2° aperture) measured 107 fl (366 cd/m2) and 262 fl (899 cd/m2) with copse and parking lot backgrounds, respectively. Although the speed limit sign had positive contrast with the parking lot background according to these measures, specular reflections from nearby car windows exceeded 84,067 fl (288,000 cd/m2). As such, contrast ratios with the parking lot could vary greatly depending on cloud cover and slight adjustments in the photometer settings.

The yellow portion of the yellow pedestrian warning sign averaged 292 fl (999 cd/m2) and 389 fl (1,331 cd/m2) with the copse and parking lot backgrounds, respectively. The fluorescent yellow portion of the fluorescent yellow-green pedestrian warning sign averaged 813 fl (2,786 cd/m2) and 1,186 fl (4,063 cd/m2) with the copse and parking lot backgrounds, respectively. Readings to the left and right of the warning signs were similar to those for the speed limit sign.

Laboratory Procedure

Participants were seated in a driving simulator cab and viewed images projected on a cylindrical screen. In each trial, participants were presented with one of the four panoramic backgrounds. The TCDs in the original photographs had been removed from the panoramas. In half of the trials, a sign was present at the location where the target TCD had been in the original photographs. The background-alone image or background-plus-sign image was presented for 0.1 s. The participant's task was to indicate whether or not the sign was present. The difficulty of this discrimination was controlled by varying where the participant was instructed to look before each scene was projected.

The FHWA's Highway Driving Simulator was used to project the static stimuli. The simulator's screen is cylindrical, with a radius of 8.9 ft (2.7 m). Directly in front of the driver, the design eye point of the simulator is 9.5 ft (2.9 m) from the screen. The stimuli were projected onto the screen by five Barco projectors, each of which displays 2,048 horizontal by 1,536 vertical pixels. Because the projection system covered 240° and the panoramic stimuli covered only about 120°, the outside projectors displayed horizontally reversed images of the left and right images of the panoramas. The reversed images could only be viewed through the side windows of the vehicle cab.

A 3-ft (0.9-m)-wide sign at a distance of 85 ft (26 m) subtends a visual angle of approximately 2°. The speed limit signs in this experiment were 4 inches (10 cm) high when projected onto the screen, which yielded a sign that subtended 2° of visual angle from top to bottom. The speed limit numerals subtended the same visual angle as would 12-inch (30.5-cm)-high numerals viewed from a distance of 85 ft (26 m). The speed limit sign used the standard FHWA series E font.(47) Measured on the diagonal, the warning signs were 4.5 inches (11.4 cm) wide on the screen and subtended a visual angle of 2.3°. Letters on these signs subtended the same visual angle as would 4.7-inch (11.9-cm) letters when viewed from a distance of 85 ft (26 m). The warning signs used the standard FHWA series C font.(47)

The average luminance of the white portion of the speed limit sign measured with a 0.2° aperture was 2.3 fl (7.8 cd/m2). The mean luminance of the black characters on the sign measured with a 0.1° aperture was 0.1 fl (0.5 cd/m2). The mean luminance of the yellow areas on the yellow warning sign measured with a 0.2° aperture was 2 fl (6.8 cd/m2), and the luminance of the black areas measured with a 0.1° aperture was 0.1 fl (0.5 cd/m2). The luminance of areas immediately to the left and right of the signs ranged from 0.1 to 0.7 fl (0.5 to 2.4 cd/m2) and averaged 0.4 fl (1.2 cd/m2).

Method of Limits

Each trial consisted of a 1-s presentation of a fixation cross in a gray field followed by the presentation of a stimulus scene for 0.1 s. The stimulus scene was followed by a gray field that remained until the next trial was initiated. The gray fields filled the entire horizontal 240° of the projection screen. Each sign (speed limit, yellow warning, and yellow-green warning) was tested in separate sessions. Each session was about 12 min in duration. Within sessions, trials were in blocks of 16. Within blocks, each of the four backgrounds was presented four times, twice with the sign present and twice with the sign absent. Within each block, the presentation order of backgrounds was random, with the restriction that each background occurred four times. Participants pressed the right key on a remote control device to indicate that the sign was present or the left key to indicate that it was absent. A press of either key initiated the next trial.

A staircase variation of the method of limits was used to arrive at the critical conspicuity detection angle for each sign with each background.(48) At the end of each block of 16 trials, the offset of the fixation cross from the sign location was incremented up or down by 3°. For each background, the direction of the increment depended on participant performance in the preceding block. If the participant was correct concerning sign presence (or absence) on all four trials with a background, then the offset for that background was incremented by +3°. If an error was made on one or more trials (e.g., the participant indicated the sign was present when it was not or vice versa), then the offset for that background was incremented by ‑3°. Testing continued in this manner until the direction of the increment had reversed a minimum of five times with each background. The critical angle for detection of the sign in each background was then computed using values of offset angles when the direction of increments had been reversed. The first reversal was not included in the computation as it is more dependent than later trials on the angle at which testing began. Thus, for each background, the critical conspicuity angle for detection was based on at least four reversals. Because testing continued on all backgrounds until the criterion of five reversals was reached for all backgrounds, scores could be based on more than four reversals. The critical angle computation averaged the angle for which an error was made on trials following offset increases and the angle for which correct responses were made on trials following offset decreases. The staircase method provided an estimate of the angle at which a participant will be correct on four trials 50 percent of the time. Table 8 provides an example of how the critical conspicuity angle was computed for one sign-background combination, a fluorescent yellow-green warning sign in an urban background. In this example, there were six reversals rather than five because testing was done in blocks of 16 trials and the participant required more trials to reach five reversals with one of the other backgrounds. The purpose of continuing to test with backgrounds for which the criterion had been reached was to keep the background uncertainty constant and thereby maintain task difficulty.

Table 8 . Example critical detection conspicuity angle computation for one participant.


Offset Angle

4 Correct Responses?

Score

Explanation

27

Yes

 

Start

30

No

 

First reversal excluded from scoring

27

No

 

 

24

Yes

24

Reversal

27

Yes

 

 

30

Yes

 

 

33

No

33

Reversal

30

Yes

30

Reversal

33

No

33

Reversal

30

No

 

 

27

No

 

 

24

Yes

24

Reversal

Mean

 

28.8

 

Blank cells indicate no score was recorded.

For two backgrounds, the offset angle for the first block of 16 trials was 27°. For the remaining two backgrounds, the starting offset angle was 45°. The starting angle for each background was counterbalanced across participants.

Laboratory Instructions

Participants were told that a fixation cross would appear on the screen before each trial. They were instructed to turn their head toward the cross and focus their eyes on it. They were told that the cross would be on the screen for about 1 s, after which a roadway scene would be presented for a fraction of a second. A picture of the sign was provided. If the sign was there, they were to press the right key on a small remote control. If they sign was not there, they were to press the left key. They were instructed to guess when unsure. They were told that the next trial would start shortly after a key was pressed. Participants were implored to "… always keep your gaze at the location of the fixation cross. Do not shift your gaze to the … sign. We are interested in finding how far away from people's gaze point a sign can be and still be detected. Shifting your gaze to the sign will defeat the purpose of this test. Therefore, keep your eyes fixated wherever the cross appears."

Results

Outdoors

The mean offset detection angles from the outdoor procedure are shown in figure 27, where the error bars represent 95 percent confidence limits of the means. The sign-background interaction was significant, F(2,11) = 9.9, p < 0.01, η2p= 0.64. The interaction was the result of the fluorescent yellow-green sign having a significantly greater detection angle than the yellow sign, regardless of background, F(1,12) = 11.3, < 0.01, η2p= 0.48, whereas the speed limit sign was only less conspicuous than the fluorescent yellow-green sign with the parking lot as a background, F(1,12) = 8.1, p = 0.02, η2p= 0.40, but not with the copse background (p = 0.15). The main effect of background was not significant (p = 0.08). The main effect of sign was significant (p = 0.04).

The abscissa of the bar graph is labeled scene background, which has two levels: trees and parking lot. The ordinate shows mean detection angle and ranges from 0 to 80 degrees. With trees as a background, speed limit signs could be detected at 70 degrees, plus or minus 7 degrees; yellow warning signs could be detected at 61 degrees, plus or minus 7 degrees; and fluorescent yellow-green warning signs could be detected at 70 degrees, plus or minus 10 degrees. With the parking lot as a background, speed limit signs could be detected at 56 degrees, plus or minus 14 degrees; yellow warning signs could be detected at 62 degrees, plus or minus 10 degrees; and fluorescent yellow-green warning signs could be detected at about 65 degrees, plus or minus 11 degrees.
Figure 27 . Graph. Mean detection offset angles measured outdoors.

Laboratory

The mean detection angles of the three signs with the four backgrounds are shown in figure 28. Error bars in the figure represent 95 percent confidence limits of the means. The maximum measureable angle was 60°. Beyond 60°, the A-pillar of the vehicle cab obstructed the screen. As a result, there was a ceiling effect that had the greatest influence on the conspicuity angle of the warning signs when presented with the suburban and copse backgrounds. In several cases, participants reached the 60° ceiling with few or no reversals. In such cases, the session was terminated before the criterion of five reversals was reached and the conspicuity angle was recorded as 60°. Despite this limitation, clear differences in detection conspicuity were obtained both for sign type and background. A reduction in standard error as a result of the ceiling is clearly shown in the confidence limits for the fluorescent yellow-green warning sign with the copse background.

This graph shows mean detection angle for three types of signs (speed limit, yellow warning, and fluorescent yellow-green warning). Detection angles are shown separately for urban, suburban, parking lot, and tree backgrounds. Speed limit means were as follows: urban, 20 degrees; suburban, 51 degrees; parking lot, 23 degrees; trees, 52 degrees. Yellow warning sign detection angles were as follows: urban, 30 degrees; suburban, 49 degrees; parking lot, 47 degrees; trees, 54 degrees. Fluorescent yellow-green sign detection angles were as follows: urban, 39 degrees; suburban, 54 degrees; parking lot, 48 degrees; trees, 58 degrees.
Figure 28 . Graph. Mean critical detection conspicuity angles measured in the laboratory.

Using repeated measures analysis of variance, the sign by background interaction was significant, F(6,6) = 11.3, p < 0.005,η2p= 0.92, as were the main effects of sign type, F(2,10) = 47.7, p < 0.001,η2p= 0.91, and background, F(3,9) = 116.1, p < 0.001, η2p= 0.98. These effects are shown in figure 28. The mean detection angle with the copse background (54°) was significantly greater than those of the urban (29°) and parking lot (39°) backgrounds but was not significantly different from the mean detection angle with the suburban background (51°). The mean detection angle of the fluorescent yellow-green sign (50°) was significantly greater than that of the speed limit sign (36°), F(1,11) = 74.5, p < 0.001, η2p=0.87, and that of the yellow warning sign (45°), F(1,11) = 5.2, p = 0.04, η2p= 0.80.

The interaction of sign and background can be traced to the comparison of the speed limit sign with the fluorescent yellow-green sign. In those comparisons, the interaction is significant when the urban environment is compared to the copse, F(1,11) = 14.8, p < 0.01, and when the parking lot is compared with the copse, F(1,11) = 14.6, p < 0.01, but not for the comparison of the suburban environment with the copse (p = 0.46). When yellow and fluorescent yellow-green comparisons are considered, all interactions with background are non-significant (p > 0.10). With the suburban and copse backgrounds, the differences in detection angle between sign types were not significant (= 0.07), nor was there a significant difference in detection angle between those backgrounds (p = 0.09).

Discussion

Signs must be detected before they are processed. This experiment showed that, in natural environments, speed limit and warning signs can be detected at angles of 60° or more from the point of gaze. In low contrast environments, such as those in the laboratory, the detection angles were still substantial-over 50° with an uncluttered background that provided reasonable color contrast. With busy or cluttered backgrounds and little contrasting color or luminance (e.g., speed limit sign with parking lot background), the detection angle is substantially reduced. Notably, Cole and Jenkins also reported that regulatory signs were less conspicuous than other colored signs used in their study.(2)

The 20° detection angle for the speed limit sign in the laboratory was obtained when participants were actively monitoring for sign presence. In a real-world context in which drivers are not actively searching for speed limits or warnings, it is reasonable to assume that detection angles would be considerably less than those observed. Nonetheless, the findings are relevant to real-world signing. If a warning or speed limit message is important to communicate, then consideration of factors that maximize the probability of detection are also important. The present findings suggest that the environment around signs affects their detection conspicuity. The speed limit sign stood out against backgrounds of leafy green trees but was much less conspicuous against a background of cars, pavement, advertising signs, and other objects. With light-colored surrounds, strong consideration should be given to making speed limit signs, and perhaps other regulatory signs, more conspicuous. The results do not suggest how this should be done, but two common approaches are to make the signs larger or to use a conspicuous contrasting border. Given the superior detection conspicuity of fluorescent yellow-green signs, perhaps fluorescent yellow should be considered as a candidate for enhancing speed limit sign conspicuity. The 2009 MUTCD provides for a yellow notice plaque (W16-18P) that might increase speed limit sign conspicuity.(36) Given the study's results, it would seem that this plaque should be seriously considered wherever the surrounding environment provides poor contrast with the speed limit sign.

The fluorescent yellow-green warning sign was less sensitive to background clutter perhaps because its color contrasted with all the backgrounds used. Had nearby elements in the scenes been similar to the yellow-green color, the sign may have suffered detection degradation. It should be noted that because they fluoresce in natural light, fluorescent yellow-green signs have greater luminance than standard yellow signs. This may have contributed to the slight advantage the fluorescent sign had in the outdoor detection test. However, in the laboratory the detection test, fluorescence was not a factor. Nonetheless, the yellow-green color was 14 percent greater in luminance than the standard yellow color.

Color is an important visual property, even in peripheral vision. Although it has been reported that color perception is absent beyond 40°, more recent research has shown that, for relatively large targets, opponent cone color perception is retained to at least 50°.(49,50)

This study demonstrated that the angle away from the point of gaze is an important consideration in the detection of signs, especially black-on-white regulatory signs in light-colored and cluttered background environments. This has important implications for the placement of these signs. On wide roadways, such as 8- or 10-lane interstates, speed limit signs 12 ft (3.7 m) or more to the right of the travel way may not attract drivers' attention. Near intersections, a common placement for speed limit signs and roadside regulatory signs may be far from the gaze point of drivers making turning movements. In the case of wide roadways, conspicuity enhancement should be considered. In the case of intersections, consideration should be given to midblock placement of speed limit signs and conspicuity enhancement should be considered for turn restriction signage.

The specific objectives of the conspicuity detection experiment were to determine how close to the direction of gaze signs need to be for their presence to be detected and whether laboratory methods are sufficient for assessing conspicuity angles. The results suggest when the observer's only task is to detect sign presence, speed limit and warning signs are detectable at any angle a driver is likely to encounter them. The exceptions to this may be signs placed at intersections or other locations where the driver's focus of attention is more than 20-40° away from the location of the sign and where other the demands on attention are high. The results also suggest that Wertheim's conspicuity angle measurement techniques can be adapted to the laboratory and provide interpretable results even when the contrast ratio between signs and surrounds is orders of magnitude less than in a natural environment.(7) Shape and color of backgrounds, rather than the magnitude of contrast ratios, seem to be determining factors in TCD detectability, at least with the ranges of contrast examined in this study.

It was found that signs are detectable at large angles of offset from the direction of gaze. However, in the on-road field study, drivers were reporting the content of signs that glance data indicated they had not looked at directly. To confirm that drivers can read signs they do not look at directly, the laboratory experiment reported next was performed. In the following experiment, the off-axis angle at which signs can be read was explored.

ResearchFHWA
FHWA
United States Department of Transportation - Federal Highway Administration