Search

This project used the second Strategic Highway Research Program (SHRP 2) naturalistic driving study (NDS) data as an alternate data source to police-reported crash data to better understand roadway object crashes. The objectives included determining crash causation, recommending strategies for crash prevention, and understanding the implications for highly automated vehicles (HAVs). Researchers addressed these objectives with a three-pronged approach: (1) a detailed engineering study of roadway object events to identify and quantify effects of a large number of relevant variables; (2) a machine-vision-oriented study to document the implications of roadway object events on machine vision performance; and (3) a detailed case study analysis of representative roadway object events to provide further qualitative results on how and why roadway object crashes occur and what potential actions can prevent such events effectively.

Nighttime crashes at intersections are a major traffic safety concern in the United States. Although providing lighting at intersections has proved to be a successful intervention against night crashes, current approaches to designing lighting at intersections are relatively simplistic, based on recommending light levels. These light levels stem from research that evaluated the effect of intersection lighting on night crashes, which does not account for the role of a driver’s visual performance or the effects of vehicle headlamps. For effective lighting design at intersections, empirical research is required to evaluate the effects of intersection lighting design on a driver’s visual performance as well as perceived visibility and glare. The current study had two goals. The first was to quantify visual performance in three lighting configurations (illuminating the intersection box, approach, or both). The second was to determine what lighting levels within each lighting configuration support the best visual performance. The study involved a target detection task, completed at night on a realistic roadway intersection. Twenty-four participants completed the study, with equal numbers of younger (18–35 years) and older (65+) individuals. Illuminating the intersection box led to superior visual performance, as indicated by longer target detection distances, fewer missed targets, and more targets identified within a safe stopping distance. For this lighting configuration, visual performance plateaued between an illuminance level of 8 and 12 lux. Visual performance was inferior in lighting configurations in which only the approach to the intersection or both the approach to the intersection and the intersection box were illuminated, and there was not consistent plateauing of visual performance in either condition. Increased performance with box lighting was likely due largely to the rendering of targets involved. Visual performance was reduced among older participants, though age-related differences were consistent across lighting configurations. These results have important implications for the design of intersection lighting at isolated or rural intersections. Specifically, results indicated that illuminating the intersection box is an effective strategy to increase nighttime visual performance for a wider range of driver ages and could also be an energy-efficient solution.

This study sought to determine if different types of roadway lighting performed differently in rain or fog. The performance of the lighting was determined by participants’ ability to detect different objects along the shoulder of the road as they drove an experimental vehicle through simulated rain and fog at night on the Virginia Smart Roads Highway. Twenty-seven participants took part in this study, which consisted of three sessions: one for consenting and screening, one for performing the study in clear weather, and one for performing the study in rain and fog conditions.. As participants drove along the Smart Roads Highway, they looked for and verbally identified two types of objects that appeared on the right shoulder of the road. These included pedestrians wearing red, blue, or gray clothing, or small 7-inch square wooden targets painted red, blue, or gray. Participants also identified the color of the object. Participants performed these tasks under three different types of roadway lighting: traditional high-pressure sodium (HPS) lamps and two types of light-emitting diode (LED) lamps with different color temperatures (3500K and 6000K). Presentation orders of the lights and objects were counter-balanced to reduce learning effects. The performance of each light was determined by the distance at which participants could identify objects (detection distance) and the distance at which they could recognize the color of the objects (recognition distance). For each object and weather condition, a 2 (age) × 3 (lighting) × 3 (color) analysis of variance (ANOVA) was conducted with an alpha of 0.05. No significant difference was found among light types on participants’ ability to detect pedestrians in any weather condition. Results showed that all three light types performed equally for the detection of pedestrians in all weather conditions and for the detection of targets in clear and fog conditions. A significant difference was only found for the detection of targets in rain. However, there was no clear best performer as each light type performed well for the detection of some colors of targets and poorly for others. On average, detection distances for targets in the rain were approximately 10 m longer under the LED lights compared to the HPS.

The goal of this study was to develop and demonstrate the effectiveness of a novel bicycle data acquisition system (bikeDAS) to extend the Virginia Tech Transportation Institute’s expertise in instrumentation and naturalistic data collection into the realm of bicycles and other fully or partly human-powered vehicles. The methodology incorporated the identification of a series of design considerations for effective bicycle instrumentation, including weight and balance, power, data collection capability, and mount location and design. These were addressed in a series of prototype iterations, each of which included testing in on-road situations to determine areas for improvement. The final VTTI bikeDAS is based on a weatherproofed version of the VTTI MiniDAS. The MiniDAS was developed as a low-cost, self-contained data acquisition system (DAS) for use in automobiles and motorcycles and has been adapted for use on bicycles with the addition of a rechargeable battery pack that looks and mounts to the frame like a standard water bottle. It consists of a single main unit measuring approximately 165 mm by 133 mm that contains two video cameras and numerous sensors, including GPS and a multi-axis inertial measurement unit (IMU). A mounting system has been developed to allow flexibility of placement across different types of bicycles; the typical location is attached to the head tube, which allows an unobstructed forward view, as well as views of the rider’s face and hands, and approaching vehicles, depending on bicycle geometry. This report provides detail about the needs analysis, prototyping, development, and deployment phases, and discusses areas of future development and research.

Safety at unsignalized intersections continues to be a major concern for transportation agencies and roadway users. To improve intersection safety, this project conducted a comprehensive study of gap acceptance behaviors at unsignalized intersections using the second Strategic Highway Research Program (SHRP 2) naturalistic driving study (NDS) data. The team collected 1,170 accepted and rejected gaps/lags based on 466 NDS trips at 60 unsignalized T-intersections in Washington state and North Carolina. The project team utilized a number of data sources, including time series data measuring vehicle kinematics for the analyzed trips, forward-facing and rear-view videos for the analyzed trips, driver demographic and driving history data, the SHRP 2 Roadway Information Database, and satellite images. First, the team identified the critical gaps for a number of common scenarios using three widely accepted methods: binary logistic regression, maximum likelihood method, and probability equilibrium method. Results showed an overall critical gap of 5.3 seconds for right-turning trips and 6.2 seconds for left-turning trips. The team then went on to develop a complete understanding of the factors affecting gap acceptance decisions using logistic regression and machine learning techniques. A number of factors were identified that affect drivers’ gap acceptance decisions, including being a gap instead of a lag, presence of leading and/or following vehicles, higher volume, intersection being unskewed, and increased number of through lanes. Finally, researchers further investigated drivers’ longitudinal and lateral acceleration behaviors during turning after accepting a gap and factors affecting their turning behaviors. Overall, both left- and right-turning vehicles initially accelerated quickly after they accepted a gap, and then reduced to a lower but prolonged acceleration rate while turning to reach a desired speed. For lateral acceleration, the peak value for the left-turning profile was reached later in the turning process than for the right-turning profile.

This report evaluates an alternative method for lighting highway signs that takes advantage of their retroreflective properties. This method uses a single luminaire mounted some distance upstream of the sign, with a focused, but evenly distributed beam so that the sign always receives the same amount of illumination. To evaluate the proposed system, a human-subjects experiment was performed on the Virginia Smart Road to test two sign configurations: an overhead sign mounted on a gantry above the highway and a sign mounted at the side of the road on a horizontal curve. For the overhead sign, there were no statistical differences between the upstream lighting, traditional style lighting, and headlamp-only conditions, though there was an increase in the mean legibility distance of the sign (~14 m) when the upstream luminaire was located on the shoulder of the road at the highest luminance setting. For the horizontal curve, upstream lighting provided no benefit for sign legibility due to the ambient lighting in the vicinity, but placing the luminaire closer to the sign (approximately 20 m) resulted in more consistent luminance from a wider range of viewing angles. Although upstream lighting provided legibility distances similar to traditional sign lighting, it may offer advantages due to reduced maintenance and energy costs.

Cyclist deaths are overrepresented among traffic fatalities, and increasing cyclist conspicuity to drivers could potentially reduce cyclist deaths, particularly at night. This report describes an experiment with various commercially available bicycle visibility-enhancement systems in terms of their conspicuity to drivers during the day and at night. Visibility enhancements included a headlamp, tail lamp, spoke lights, and retroreflective clothing, including garments that highlight biomotion. The results indicate that active visibility treatments, such as bicycle-mounted lights, make cyclists more conspicuous than passive systems like retroreflective vests and biomotion bands. Flashing headlamps and tail lamps were the most conspicuous treatments during both the day and at night; fast flashing headlamps (6.7 Hz) had higher detection distances and rates during the day, and moderately fast flashing headlamps (3.4 Hz) had higher detection distances and rates at night. Spoke lights and flashing tail lamps, along with retroreflective vests, also aided cyclist visibility during the day and at night, especially for vehicles approaching intersecting cyclists. Passive retroreflective visibility treatments were most effective at night, when the vehicle was passing the cyclist from behind. However, that approach also used reflectors, so the discrete effect of passive retroreflective treatments could not be determined. This study also found that biomotion markers alone do not significantly increase cyclist conspicuity in visually complex natural environments. For most approaches, flashing lights had greater detection distances than biomotion markers, which in turn had higher detection rates than headlamps and tail lamps.

Virtual reality (VR) promises to be an effective tool to evaluate changes to the built environment that could improve safety for pedestrians. However, in order to draw actionable conclusions from VR, it is important to understand the degree to which pedestrians’ perceptions and behaviors match across real and virtual environments. In this study, participants experienced equivalent real and virtual environments and performed similar tasks in each. Tasks included the intention to cross an intersection, the estimation of the speed and distance of an approaching vehicle, and the perceived safety and risk of crossing a road. Results showed no statistical difference between the real and virtual environments for participants’ intention to cross, estimation of distances, and perceptions of safety and risk. Statistically significant differences between real and virtual environments were observed in the estimation of speed and measures of presence. These results indicate that at lower vehicle speeds (25 mph and lower) VR can be used as tool to evaluate pedestrian safety in built environments.

In 2015, approximately 8,000 intersection and intersection-related fatal crashes occurred on the nation’s highway system, resulting in more than 8,400 fatalities. That death toll represented about 24% of the traffic-related deaths across the country. Combining fatalities and injuries, intersection and intersection-related crashes represent more than 50% of the traffic-related injuries across the nation. Unsignalized intersections are of particular concern. Between 2010 and 2014, unsignalized intersections were responsible for more than 70% of the intersection and intersection-related fatalities. This report documents 83 suitable safety countermeasures that can be used at unsignalized intersections to mitigate crash risks. A number of these have potential for cost-effective, systemic implementation, including LED-enhanced Stop signs, retroreflective panels on sign posts, center line pavement markings in a median crossing, center line pavement markings on the minor road approach, and installation of intersection lighting.

Vehicle headlamps and roadway lighting are the major sources of illumination at night. These sources affect contrast-defined as the luminance difference of an object from its background-which drives visibility at night. However, the combined effect of vehicle headlamps and intersection lighting on object contrast has not been reported previously. In this study, the interactive effects of vehicle headlamps and overhead lighting on object contrast were explored based on earlier work that examined drivers' visibility under three intersection lighting designs (illuminated approach, illuminated box, and illuminated approach + box). The goals of this study were to: 1) quantify object luminance and contrast as a function of a vehicle's headlamps and its distance to an intersection using the three lighting designs; and, 2) to assess whether contrast influences visual performance and perceived visibility in a highly dynamic intersection environment. Both luminance and contrast of roadway visibility targets and a pedestrian were measured with a calibrated photometer at a realistic intersection. Both target and pedestrian contrast and luminance were substantially affected by the intersection lighting configuration, illuminance level, location at the intersection, and vehicle distance from the intersection. Objects also underwent changes in contrast polarity (positive to negative or vice-versa) as the distance between the vehicle and object changed. During these polarity transitions, objects became invisible because the contrast was zero. Negative contrast on targets was associated with higher visual performance. Within a given contrast polarity (positive vs. negative), visual performance depended on the magnitude of contrast, with higher contrast associated with higher visual performance. The relationship between pedestrian contrast and perceived visibility was complex, since pedestrians were often rendered in multiple contrasts. These findings have important implications for the lighting design of intersections and the development of nighttime pedestrian detection systems that rely on computer vision. Published version

Fog- and weather-related visibility reduction is a common cause of multiple-vehicle crashes. Large differential speeds and a tendency of vehicle operators to drive faster than is safe can lead to terrible crashes. Fog can usually be seen on traffic cameras, which are becoming more prevalent on Virginia highways as well as on highways in other states. This project studied the applicability of one approach to using machine vision to measure fog in a realistic environment simulated on the Virginia Smart Road. With the assistance of Dr. Eric Dumont, a leading visibility research from IFSTTAR in France, a machine vision algorithm was applied to video stills captured from a common traffic camera installed on the Smart Road. Machine vision algorithms were used to determine the average loss in visual detail in the scene viewed by the camera and this was used to generate an empirical model relating Meteorological Optical Range (MOR) and the camera images. The model was used to evaluate data captured on days close in time and days over the following year. Finally, the research investigated the approach’s sensitivity to preset positioning errors in the camera. The research shows that the approach has promise. However, further research and development are needed before the approach is ready for deployment.

The purpose of this project was to evaluate the effectiveness of an alternative cooperative headlighting method, dubbed 4U Lighting. A human-subjects study was conducted in which 12 participants 65 or older observed pedestrians under different lighting configurations and identified the moment when they were sure they could see a pedestrian. The participants drove a vehicle towards a static vehicle in the opposite lane around which the pedestrians were located. The distance at which participants could detect the pedestrian, termed the detection distance, was compared across lighting conditions and served as the measure of improvement in driver visual performance (visibility). Commercial connected vehicle hardware and protocols were used to communicate position between the two vehicles and to trigger the operation of the custom lighting control system. The system operated as expected and the data showed benefits to driver visual performance.

Portable light towers are a significant source of glare to motorists entering a work zone. Although existing research has evaluated the effect of light tower orientation on visibility and glare, the effects of factors like mounting height, offset distance from the roadway, and number of light towers in the work zone on visual performance and discomfort glare are not known. Understanding these relationships can help to develop illumination guidelines for work zones that can reduce glare for drivers. The goal of this project was to understand the effect of mounting height, offset distance to the roadway, and number of light towers in the work zone on drivers’ visual performance and discomfort glare. Participants drove through a realistic work zone and evaluated portable light towers with varying mounting heights, offset distances, and number of light towers. Results showed that the mounting height and offset distances play a critical role in affecting the driver’s visual performance and discomfort glare rating. Portable light towers, irrespective of wattage and lumen output, at lower than a mounting height of 20 feet and closer to the roadway (in travel lanes than in the shoulder) result in decreasing drivers’ visual performance and increasing their discomfort glare. Portable light towers should be mounted at a height of at least 20 feet, and balloon light towers with higher wattage (4,000 watts and greater) and lumen output (400,000 lumens and greater) should be located at an offset distance of at least 10 feet from the roadway.

Roundabouts increase safety, but their safety effects on vulnerable road users are not as great as the safety effects for motor-vehicle drivers. Additionally, many motorists fail to yield to pedestrians at intersections and roundabouts, possibly because drivers do not see the pedestrians, crosswalk signage, and/or pavement markings. Eye-tracking technology has been used to quantify driver eye-glance behavior in a variety of driving contexts, but has yet to be applied to drivers in roundabouts with pedestrians. The exploratory research performed for this project attempted to shed light on driver visual behavior at roundabouts, and to examine the extent that drivers looked at pedestrians, with the ultimate goal of increasing pedestrian safety at roundabouts. A quasi-naturalistic experimental study was conducted and drivers’ mean fixation durations toward pedestrians located at crosswalks were measured at two different roundabouts while making three kinds of turn maneuvers (straight through, left turn, and right turn) under day and nighttime conditions. Three important findings were evident. First, the results show that the position of the vehicle within the roundabout (approach, entry, and exit) had a significant impact on drivers’ visual behavior toward pedestrians. Drivers looked at pedestrians longer when they (drivers) were at the approach portion of the roundabout and less at the entry and exit portions of the roundabout. Second, the number of lanes at the roundabout did not significantly affect the drivers’ fixation durations toward the pedestrians. Finally, the age of the drivers and time of day did not significantly influence fixation durations toward the pedestrians at the roundabouts.

Run-off-road crashes are a major problem for rural roads. These roads tend to be unlit, and drivers may have difficulty seeing or correctly predicting the curvature of horizontal curves. This leads to vehicles entering horizontal curves at speeds that are too high, which can often lead to vehicles running off the roadway. This study was designed to examine the effectiveness of a variety of active and passive curve warning and curve delineation systems on two two-lane rural roads to determine which is the most effective at reducing vehicle speeds and assisting lane-keeping. The study consisted of a human-factors study, as well as an observational study. There were nine curves examined in the study on two road sections in Southwest Virginia. The human-factors study included participants whose speed and lane position were tracked as they drove through eight curves, both before and after new treatments were installed in each of the eight curves. The observational study examined the speed and lane position of traffic on all the curves before and after the installation of the new treatments. The observational study included a curve on a road near the primary study section. The results of the study were mixed, with every tested system leading to some reductions in speed or encroachments at some parts of the curve while also leading to increases in the same values at other parts of the curve. No clear difference was discovered between passive and active systems or between delineation and warning systems. The study recommends that in addition to a safety assessment, specific curve characteristics and budget should be the main considerations in the selection of a treatment for a curve. Virginia Department of Transportation Federal Highway Administration Contract or Grant No. 103652 Final Contract

In recent years, the complexity of the evaluation metrics of roadway lighting has continually increased. In order to measure the effectiveness of a roadway lighting system in situ, a mobile measurement system has been developed to provide evaluations of illuminance, luminance, spectral content, glare, human visibility performance, and vehicle condition. This system, the Roadway Lighting Mobile Measurement System (RLMMS), has been developed to be transportable to anywhere in the world and mounted on any vehicle. The RLMMS consists of five illuminance meters: four measure the centerline and wheel path illuminance of the vehicle and the fifth measures the illuminance at the eye of the driver. The system also uses two cameras: the first a calibrated luminance camera and the second a color camera for scene evaluation. The RLMMS is able to tie into the internal vehicle network to obtain speed and distance measurements. Additional tools, such as a spectroradiometer and an eye tracker, have also been incorporated into the system. Finally, as a method of evaluating human response, the RLMMS includes buttons that allow users to place response markers in the data stream. The repeatability and reproducibility of the system was evaluated, and comparisons were made to the traditional handheld method of measuring illuminance. Results indicate that the system’s measurement of illuminance may be considered acceptable. The entire system has been programmed in LabView and allows for a variety of configurations of data collection. The system has been utilized in several research projects, including some in Anchorage, Alaska; Honolulu, Hawaii; and San Diego, California.

This project explored the feasibility of using Second Strategic Highway Research Program (SHRP 2) data, including the Roadway Information Database (RID), to evaluate the effectiveness of roadway safety improvements where traditional crash data are limited. The research team conducted two case studies based on naturalistic driving study (NDS) data from 200 trips. The two case studies evaluated the safety effects of (1) a paving project with newly installed pavement and markings, and (2) a median barrier replacement project with a newly installed and restored concrete median. A number of safety surrogate measures were used to develop a comprehensive understanding of how driver behavior changed with and without the safety treatment. The results from both case studies indicated that the roadway improvements had an impact on driver safety behavior, as indicated through the surrogate safety measures of speed, lateral and longitudinal accelerations, lane deviation, and car-following behavior. The two case studies illustrate two different methods for studying the effectiveness of roadway improvements on safety. The paving project case study compared driver behavior data collected at the project site after the roadway improvement with data from an adjacent site with similar roadway conditions but without the pavement improvement. The median barrier project case study compared data on the same segment of road before and after the improvement project. The two different methods illustrate the flexibility available with SHRP 2 safety data. In addition to the case studies, the research team also assessed the availability, suitability, and limitations of SHRP 2 and RID data for evaluating the safety impact of roadway improvements.

Heavy fog presents a significant safety hazard to drivers by reducing their ability to see the roadway and other vehicles. Even so, drivers often fail to adjust their speed to account for the reduced visibility, resulting in stopping distances that exceed visibility distance. Active delineators, or in-pavement light emitting diode (LED) markers, are an emerging technology which can be used to assist drivers in traveling through fog-prone areas by marking road or lane boundaries. However, there is a lack of research indicating how the presence of active delineators might also affect a driver’s behavior. This study sought to examine how the presence of active delineators in fog might affect drivers’ speed. Three lighting patterns and two brightness levels were tested. In daytime conditions, the delineators had no effect on speed. During nighttime conditions, participants often drove faster when the active delineators were present. Participants felt that the active delineators were helpful for navigating through the fog in both daytime and nighttime conditions, but preferred the higher brightness settings in the daytime.

This project evaluated current nighttime work zone lighting practices for limited-access highways and primary routes in Virginia through (1) an on-site evaluation of lighting levels in work zones; (2) an illuminance characterization of various commercially available light towers; and (3) a human factors evaluation of those light towers and developed effective nighttime work zone lighting requirements for Virginia. The majority of the static nighttime work zones used metal halide portable light towers. Mobile operations such as milling and paving used equipment-mounted balloon lights and LEDs. Horizontal illuminance levels in the work zones were affected by the number of light towers, locations of the light towers, and number of traffic lanes in the work zone. The measured horizontal illuminance levels in the work zones were much higher than recommended levels. Milling and paving operations that used equipment-mounted lights had lower illuminance levels than operations that used portable light towers. Vertical illuminance levels in the traffic lane were significantly affected by the aiming of the luminaires on the portable light towers. Luminaires aimed into the traffic travel lane produced higher vertical illuminance levels, which can result in disability and discomfort glare and consequently reduce visibility. The visual performance of drivers in a work zone can be influenced by the type and orientation of the light tower. An orientation aimed toward the driver resulted in lowering drivers’ visual performance, both objectively and subjectively. This decrease in visual performance could be attributed to higher vertical illuminance. To increase the drivers’ visual performance and reduce glare in the work zone, efforts should be taken to aim the light towers in an active nighttime work zone away from the direction of traffic or perpendicular to it. In these orientations, all the three light towers tested had similar visual performance measures. The increase in the mean vertical illuminance level in the critical range is associated with higher perceived ratings of glare. Results showed that the mean vertical illuminance in the distance range of 260 to 65 ft to the light tower could be used as an objective measure of glare. A mean vertical illuminance of less than 17 lux resulted in lower perceived glare ratings. Results also indicated that light towers should be oriented so that the angle between the beam axis and driver line-of-sight axis is always greater than or equal to 90 degrees. Finally, a draft specification outline including a plan for on-site lighting evaluation of a work zone is presented. Virginia Department of Transportation Federal Highway Administration Final Contract Contract or Grant No.: 105582

Under the sponsorship of the National Surface Transportation Safety Center for Excellence (NSTSCE), this research studied the relationship between lighting level and the night-to-day (ND) crash ratio at rural intersections in the state of Virginia. Most existing research on intersection lighting indicates that the presence of lighting reduces night crashes. This study aimed to quantify the effect of lighting level and lighting quality on ND crash ratios at rural intersections. Lighting data were collected from 131 rural intersections in Virginia, and crash data for the intersections were obtained from the Virginia Department of Transportation (VDOT). Lighting data were collected using a Roadway Lighting Mobile Measurement System (RLMMS). Out of the 131 intersections, data from 99 intersections were used for the comparative analysis. Data from 32 intersections could not be used because of issues with lighting data (e.g., Global Positioning System, illuminance data dropouts). Negative binomial regression was used to model the crash and lighting data. The results showed that increasing the average horizontal illuminance at all the intersections (both lighted and unlighted) by one unit (1 lux) decreased the ND crash ratio by 7%. For the lighted intersections, the same increase in average horizontal illuminance decreased the ND crash ratio by 9%. The largest decrease in the ND crash ratio was for unlighted intersections, where a 1-lux increase in the average horizontal illuminance decreased the ND crash ratio by 21%. The average roadway luminance also had negative parameter estimates, indicating that an increase in average roadway luminance results in a lower ND crash ratio. Stop-controlled intersections had smaller ND crash ratios compared to signalized intersections. Intersections with a posted speed limit of less than or equal to 40 mph had lower ND crash ratios compared to intersections with a posted speed limit higher than 40 mph. Results also showed that most lighting levels at most rural intersections did not meet the standards recommended by the Illuminating Engineering Society of North America (IESNA).

Under the sponsorship of the National Surface Transportation Safety Center for Excellence (NSTSCE), a research team at the Virginia Tech Transportation Institute (VTTI) developed a color camera system that can collect naturalistic video data with accurate color rendering. Photometric devices can accurately measure color but cannot record the video data necessary for understanding visibility in dynamic environments like nighttime driving. Video recorders can take video data but are inaccurate with respect to color measurement. To measure color and its effects on visibility in naturalistic settings, a color camera system was developed that can record video data with color rendering similar to what humans perceive. This system includes a calibrated color camera and image analysis software. The camera system was selected and calibrated in different lighting scenarios using a standard color chart. Custom MATLAB programs were used for this calibration. These calibration files were compared for color-rendering accuracy, and the best file, based on calibration in daylight, was selected for further analysis. Researchers then used the color camera system, calibrated with the daylight file, to collect data in a variety of naturalistic settings. The color space coordinates from the color camera’s images were compared with those taken with a color meter and a digital photometer. When the camera was calibrated to daylight, it produced the most-accurate images, even when taking images in artificial lighting. Shorter exposure times produced darker images but more-accurate color space coordinates. After calibration and exposure adjustment, the color camera’s chromaticity coordinates (x, y) had about 10% error with respect to the color meter. The color camera’s luminance value (Y) had less than 5% error with respect to the color meter. The calibration file produced can be used with multiple cameras. A new image analysis method was developed. It and its accompanying custom MATLAB programs allow researchers to select portions of an image and analyze their three-dimensional color space coordinates. This capability will be useful in future work; for example, comparing photometric equipment, and analyzing naturalistic video data.

This article is part of the Proceedings of the 6th Annual International Cycling Safety Conference held in Davis, California, USA on September 20th through 23rd in the year 2017.Paper ID: 91