Traffic Signal/Road Marking Detection andProcessing

By Shannon Crouch

Humans have visual cues that they naturally use to perceivetheir motion through the environment. There are numerous humanfactors that are associated with being able to navigate a vehiclesafely while adhering to signal lights, signs and other traffic roadmarkings. There needs to be vast improvement in the way thatinformation is presented to drivers for many reasons. One example isthat the placing of the sun during a particular part of the dayprohibits one from distinguishing whether a traffic light is green orred. The elderly have difficulties reading certain signs ordifferentiating among the color of signal lights, even with theirglasses on. Most everyone has less visual acuity at night and certainroad markings or traffic lights are harder to perceive. Color blindpeople can’t always tell the difference between a red light and agreen light, and they may also have a hard time seeing brake lights,blinkers, etc. All of the situations mentioned above can lead toserious fatal accidents. Extensive research in this area has beenconducted for many years and is still going on today.

Driving is a complex task which involves a variety ofinteractive parallel and serial processes that depend on variousvisual functions. Vision plays a vital role in safe, proficientdriving even though there are other sensory and motor systems thatare used in driving. It has been estimated that 90% of theinformation used in driving is visual and that visual informationalone might be sufficient for safe driving (Fox, 1997). Beyond thesegeneral statements, it has been difficult to determine what specificvisual skills are used for driving that are important.

Safe and efficient driving is “a matter of perceptual-motorsensitivity to perceptual laws of locomotion in a spatiotemporalfield” (Fox, 1997). Specifically, the driver must be aware of twofields: (1) the field of safe travel and (2) the minimum stoppingzone. The field of safe travel refers to a field surrounded by actualand potential obstacles to locomotion. The minimum stopping zonerefers to a field that is determined by variables like speed,visual/road conditions, etc. Car crashes occur when another car orstationary object is not perceived accurately. In order to guide avehicle, the driver must (1) abstract important sensory informationfrom the environment, (2) process this information accordingly tocomplete visuospatial tasks and (3) act accordingly to achieve taskgoals. The driver must maintain awareness of his/her path and ofnon-moving and moving objects.

To safely navigate a vehicle in a consistent manner with thedriver’s goals is dependent upon a current assessment of the alteringsituation and also requires stable attentional mechanisms to keep upwith ongoing environmental information and evaluate relevantinformation.

The goal of navigation is to achieve movement through space andbecause large movements are typically more significant than smallone, it is apparent that the task must be fundamentally stored in ourunderstanding of space. We need to understand how people judge theirmovement through space and how displays may be designed to facilitatethis judgement. We also need to consider how navigational performancecan be supported through the design of maps and instructions(Wickens, 1992).

The human factors problems associated with mas and navigationshould be self-evident to anyone who has ever come across thefollowing situations: (1) driving through a confusing series ofintersections; (2) following directions on how to get somewhere andmissing a turn; and (3) having known the area of a place and thenbecome unable to locate one’s position on the map because thesurrounding landmarks are all unfamiliar (Wickens, 1992). We need tobe able to navigate from one location to another, using landmarks orother features (visual) to trigger the decisions as to whichdirection to go at a given intersection.

Reaction time (RT) is a human factor that affects signalinformation processing. RT is lengthened as a set of stimuli are madeless discriminable from one another. These discriminabilitydifficulties can be minimized by deleting shared and repeatedfeatures where possible. RT is lengthened as the discriminabilitybetween the responses is reduced (Wickens, 1992).

As we move through an environment in an automobile, ourjudgements of the direction and speed with which we are moving dependon information distributed across the visual field, not just in thearea of foreal vision, but in the peripheral vison as well. Althoughperipheral vision is not highly effective for recognizing objects, itis more proficient at conveying information about motion andorientation (Wickens, 1992).

A potential bias in human perception occurs because oursubjective perception of speed is heavily determined by global opticflow (the velocity of points traveling across the display surface andthe retina as we navigate). We feel as if we are traveling at agreater speed in a sports car than a bus, in part because the sportscar is lower to the ground. The global optical flow can be affectedby the density of texture over which a vehicle passes. As thisdensity increases (texture is finer), the global optical flowincreases and the driver perceives a faster velocity (Denton,1980).

The issues of whether different tasks are served differently bymore or less integrated displays is represented in the proximitycompatibility principle (Andre & Wickens, 1990), which statesthat “to the extent that information sources must be integrated,there will be a benefit to presenting those dimensions in anintegrated format.” It should also be noted that two items on acluttered display will be more easily integrated or compared if theyshare the same color (different from the clutter), but the sharedidentity of color may disrupt the ability to focus attention on onewhile ignoring the other (Andre & Wickens, 1990). A unique colorcode will help this focusing process, just as it disrupts theintegration process. Cluttering of signs and signals can causeconfusion. A study conducted by Culler, Holahan and Wilcox (1978)found that the ability to locate and respond to a stop sign in acluttered display is directly inhibited by the proximity of otherirrelevant signs in the field of view.

The proximity compatibility principle also applies to spatialdistance in a cluttered display. Two pieces of information that needto be integrated on a cluttered display should be placed in closespatial proximity, as long as this proximity does not also move themcloser to irrelevant clutter (Andre & Wickens, 1990).

Color blind individuals cannot differentiate among colors,which makes it difficult to perceive traffic signals of any kind.Color blind people are technically termed dichromats, meaning,“two-colors.” Dichromats can see only two hues whereas people withnormal vision can distinguish over one hundred hues. Most dichromatscan see blue and yellow; for them, there is a region of the spectrumbetween blue and green which has no color and appears gray or white.All colors appear to them as combinations of black, white and graywith either yellow or with blue. Many dichromats will claim that theycan easily tell the red from green in a traffic signal. In the UnitedStates, the go-light is blue-green so that color blind people candistinguish it from red. Even though they can discriminate betweenthe two, the go-light falls into the colorless region of the spectrumfor many, and it is indistinguishable from the color of automobileheadlights at night. They have become accustomed to using brightnessand saturation differences as cues and they have learned to callthese differences “colors” (Neitz, 1997).

Another way color can affect visual processing is whenisoluminant displays are used. leClerc, Malbert, Montagnon, &Troscianko (1991) revealed that certain color gradients (atisoluminance) markedly affected the perceived depth. A gradient insaturation (e.g. red-to-gray) was particularly effective in allowingdepth perception, while a red-green hue gradient had no effect onperceived slant. Their data suggests that while color can encodedepth, its contribution is contingent on the presence of texturecues. This contingency implies strong links between texture and colorprocessing in human vision.

There have been several proposed solutions to theabove-mentioned problems. To reduce the speed of individuals invehicles, since speeding plays a major role in traffic accidents, theAAA Foundation for Traffic Safety is planning a test series usingconverging patterns of stripes on the road to give drivers theillusion of excessive speed; this should, in turn, cause them toreduce their speed. The only problem foreseen with this method isthat varying speeds can themselves pose problems, since feweraccidents occur when all traffic drives at the same speed (ScienceNews, 1996).

Denton (1980) did a similar study on the previously mentionedresearch. He exploited the characteristic of the perceptualexperience, when the global optical flow increases and the driverperceives a faster velocity, in an ingenious application ofperceptual research to highway safety. His concern was withautomobile drivers in Great Britain who approached traffic circles atan excessive rate of speed. Denton’s solution was to minimize thespacing between road marks at a gradual and continuous rate as thedistance to the stop point minimized. A driver who may be going at anexcessive speed would see the global optical flow as increasing. Thiswould cause the driver to compensate by imposing a more appropriatedegree of braking or slowing down (Wickens, 1992). Results of hisstudy showed significantly slower speed following introduction of themarkers and the rate of fatal accidents was greatly decreased.

Another current research project being conducted involvesdynamic driving situations by using a version of the flickertechnique developed by Dr. Rensink of Cambridge Basic Research Centerfor studying the ability of people to perceive changes in staticscenes. An advantage of the flicker technique over others is thatawareness is probed while the subject is in the situation and notafterwards. The repeating flash can be somewhat annoying, but it doesnot affect the normal perception of a situation as it unfolds. Motionperception remains normal between flashes; this would create a“quasi-realistic but totally safe and controlled driving environment”(Fox, 1997).

This area of applied research in perception deserves furtherresearch. Our environment needs to be as safe as possible for allhumans, with visual handicaps or not, to be able to drive and processall signals and markings efficiently.


Andre, A.D., & Wickens, C.D. (1990). Proximitycompatibility and information display: Effects of color, space andobjectness of information integration. Human Factors, 32,61-77.

Culler, R.E., Holahan, C.J., & Wilcox, B.L. (1978). Effectsof visual distraction on reaction time in a simulated trafficenvironment. Human Factors, 20, 409-413.

Denton, G.G. (1980). The influence of visual pattern onperceived speed. Perception, 9, 393-402.

Fox, C.R. (1997). Reducing accident rates among elderlydrivers. 14th Biennial Eye Research Seminar, 11-12.

Illusions: The route to safer roads? (1996, July 13).Science News, 150, 31.

leClerc, J. Malbert, E., Montagnon, R., & Troscianko.(1991). The role of color as a monocular depth cue. VisionResearch, 31, (11), 1923-1930.

Neitz, M. (1997). Society and color blindness: A view in needof correction. 14th Biennial Eye Research Seminar, 64-66.

Wickens, C.D. (1992). Engineering psychology and humanperformance (2nd Ed.) New York: Harper Collins.

Return to Service Learning 1998Frontpage.