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-04-145
Date: December 2005

Enhanced Night Visibility Series, Volume XIV: Phase III—Study 2: Comparison of Near Infrared, Far Infrared, and Halogen Headlamps on Object Detection in Nighttime Rain

PDF Version (1.52 MB)

PDF files can be viewed with the Acrobat® Reader®

Alternate Text

Figure 1. Photo. Headlamp testing rack. The photo shows two halogen low beam headlamps attached to the front of an experimental SUV on a modular light rack. Back to Figure 1.

Figure 2. Photo. Object: pedestrian, blue denim clothing, left (BlueLF). This daylight photo shows a pedestrian wearing blue denim scrubs standing on the left side of the road as viewed from the participant vehicle. The pedestrian is standing just outside the far lane boundary on a straight segment of roadway. The pedestrian is standing with arms down to the side and facing the oncoming test vehicle. Back to Figure 2.

Figure 3. Photo. Object: pedestrian, blue denim clothing, right (BlueRT). This daylight photo shows a pedestrian wearing blue denim scrubs standing on the right side of the road as viewed from the participant vehicle. The pedestrian is standing just to the right of the participant’s right-hand lane boundary on a straight segment of roadway. The pedestrian is standing with arms down to the side and facing the oncoming test vehicle. Back to Figure 3.

Figure 4. Photo. Object: pedestrian in left turn, left side (LFtrnLF). This daylight photo shows a pedestrian wearing blue denim scrubs standing in a left-hand curve. The pedestrian is standing on the left side of the road just outside the far lane boundary as viewed from the participant vehicle. The pedestrian is standing with arms down to the side and facing the oncoming test vehicle. Back to Figure 4.

Figure 5. Photo. Object: pedestrian in left turn, right side (LFtrnRT). This daylight photo shows a pedestrian wearing blue denim scrubs standing in a left-hand curve. The pedestrian is standing on the right side of the road just to the right of right-hand lane boundary as viewed from the participant vehicle. The pedestrian is standing with arms down to the side and facing the oncoming test vehicle.
Back to Figure 5.

Figure 6. Photo. Object: pedestrian in right turn, left side (RTtrnLF). This daylight photo shows a pedestrian wearing blue denim scrubs standing in a right-hand curve. The pedestrian is standing on the left side of the road as viewed from the participant vehicle just outside the far lane boundary. The pedestrian is standing with arms down to the side and facing the oncoming test vehicle. Back to Figure 6.

Figure 7. Photo. Object: pedestrian in right turn, right side (RTtrnRT). This daylight photo shows a pedestrian wearing blue denim scrubs standing in a right-hand curve. The pedestrian is standing on the right side of the road as viewed from the participant vehicle just to the right of the right-hand lane boundary. The pedestrian is standing with arms down to the side and facing the oncoming test vehicle.
Back to Figure 7.

Figure 8. Photo. Object: dynamic pedestrian (PedDyno). This daylight photo shows a pedestrian walking across the roadway from the center line to the right lane boundary in front of the participant vehicle, face turned toward the participant’s vehicle, presenting a right profile of the body. Back to Figure 8.

Figure 9. Photo. Object: tire tread. This daylight photo shows a tire tread centered on the right boundary line of the participant’s lane of travel. Back to Figure 9.

Figure 10. Diagram. Likert-type subjective rating scale. This diagram shows a seven-point Likert-type straight-line scale from 1 (which equals Strongly Agree) on the left to 7 (which equals Strongly Disagree) on the right. Back to Figure 10.

Figure 11. Photo. Data collection display screen. The photo shows a sample display screen used to record the data collected during an onroad trial. The top section displays general trial information, including the experimental setup of a visual enhancement system, condition, success code, target order, day, detection and recognition code, input file, event mark code, and infrared button code. The middle section includes target order for the current run, location of the upcoming target, and running values for distance and speed. The bottom section allows the experimenter to modify the data collection process. Back to Figure 11.

Figure 12. Photo. Smart Road. This daylight photo shows an aerial view of a section of the Virginia Smart Road. The section of road is fairly straight with one gradual curve. The photo shows a rural area with no structures in the immediate vicinity. Back to Figure 12.

Figure 13. Diagram. Roadway layout. Smart Road layout with object locations. The diagram shows the entire course of the Smart Road driven by participants during the study. The road is depicted as being fairly straight with one gradual curve up toward the top. There is a turnaround loop at each end of the road. The top turnaround is in the upper left corner of the diagram, and the bottom turnaround is in the lower right corner, indicating that there is a grade. A bracket above the diagram indicates approximately half of the roadway where adverse weather conditions were generated. Four arrows along the side of the road indicate some of the locations where objects were presented. Traveling from the top to the bottom of the road, an arrow indicates one location on the left side of the road following the top turn (about a fifth of the way down the road), two locations (one on the left and one on the right) about another fifth of the way farther down the road, and one location on the right about halfway down the road. Back to Figure 13.

Figure 14. Bar graph. Tire detection and recognition distances. The graph, titled "Tire," shows the Y-axis indicates mean distance in feet and the X-axis indicates the four vision enhancement systems evaluated in the study. There are two bars for each VES, one for detection and one for recognition. Standard error bars are also provided for each graph bar. Letters above each bar note the groupings for the means (means with the same letter are not significantly different) based on the Student-Newman-Keuls results. Capital letters indicate detection groupings and lowercase letters indicate recognition groupings. The range of detection distances for the tire is from 80 feet for the NIR 1 system to 93 feet for the NIR 2 system. The range of recognition distances for the tire is from 71 feet for the NIR 1 system to 89 feet for the HLB system. The standard error bars are fairly comparable between the vision enhancement systems. There are no significantly different distances among any of the vision enhancement systems for detection and recognition of the tire. Back to Figure 14.

Figure 15. Bar graph. Blue pedestrian on straight: left and right side detection and recognition distances. The graph, titled "Pedestrian, Denim Clothing Left Side and Right Side," shows the Y-axis indicates mean distance from 0 feet to 400 feet and the X-axis indicates the four vision enhancement systems evaluated in the study. There are two bars for each VES, one for detection and one for recognition. The left half of the graph presents data for detection and recognition of the BlueLF (blue left) object with two bars for each vision enhancement system. The right half of the graph presents data for detection and recognition of the BlueRT (blue right) object for each vision enhancement system. Standard error bars are provided for each graph bar. Letters above each bar note the groupings for the means (means with the same letter are not significantly different) based on the Student-Newman-Keuls results. Capital letters indicate detection groupings, and lowercase letters indicate recognition groupings. The range of detection distances for the BlueLF is from 155 feet for the FIR system to 269 feet for the NIR 1 system. The range of recognition distances for the BlueLF is from 134 feet for the FIR system to 257 feet for the NIR 1 system. The standard error bars for detection and recognition of the BlueLF are smallest for FIR and HLB means and largest for NIR 2 means. The mean detection distances of the BlueLF for NIR 1 and NIR 2 are significantly higher than the corresponding distances for FIR and HLB. The mean recognition distance of the BlueLF for NIR 1 is significantly higher than corresponding distances for the other vision enhancement systems. The mean recognition distance of the BlueLF for NIR 2 is significantly higher than the corresponding distance for FIR. The range of detection distances for the BlueRT is from 172 feet for the FIR system to 251 feet for the NIR 1 system. The range of recognition distances for the BlueRT is from 152 feet for the FIR system to 245 feet for the NIR 1 system. The standard error bars for detection and recognition of the BlueRT are comparable between VESs. The mean detection and recognition distances of the BlueRT for NIR 1 and NIR 2 are significantly higher than corresponding distances for FIR and HLB. Back to Figure 15.

Figure 16. Bar graph. Dynamic pedestrian on straight: detection and recognition distances. On this bar graph, the Y-axis indicates mean distance in feet. The X-axis indicates the four vision enhancement systems evaluated in the study. There are two bars for each vision enhancement system, one for detection and one for recognition. Standard error bars are also provided for each graph bar. Letters above each bar note the groupings for the means (means with the same letter are not significantly different) based on the Student-Newman-Keuls results. Capital letters indicate detection groupings, and lowercase letters indicate recognition groupings. The range of detection distances for the dynamic pedestrian is from 169 feet for the FIR system to 232 feet for the NIR 1 system. The range of recognition distances for the dynamic pedestrian is from 163 feet for the FIR system to 228 feet for the NIR 1 system. The standard error bars for detection of the dynamic pedestrian are smallest for NIR 2 and largest for NIR 1. The standard error bars for recognition of the dynamic pedestrian are smallest for FIR and largest for NIR 1. The mean detection and recognition distances of the dynamic pedestrian for NIR 1 are significantly higher than corresponding distances for FIR. Back to Figure 16.

Figure 17. Bar graph. Blue pedestrian in left turn: left and right side detection and recognition distances. The graph is titled "Pedestrians in Left Turn (1,250-meter (4,140-feet) radius) Left Side and Right Side." The Y-axis indicates mean distance in feet. The X-axis indicates the four vision enhancement systems evaluated in the study. The left half of the graph presents data for detection and recognition of the LFtrnLF (Left Turn Left) object. Each vision enhancement system is represented by two bars, one for detection and one for recognition. The right half of the graph presents data for detection and recognition of the LFtrnRT (Left Turn Right) object. Each vision enhancement system is represented by two bars, one for detection and one for recognition. Standard error bars are also provided for each graph bar. Letters above each bar note the groupings for the means (means with the same letter are not significantly different) based on the Student-Newman-Keuls results. Capital letters indicate detection groupings, and lowercase letters indicate recognition groupings. The range of detection distances for the LFtrnLF is from 127 feet for the FIR system to 254 feet for the NIR 1 system. The range of recognition distances for the LFtrnLF is from 105 feet for the FIR system to 249 feet for the NIR 1 system. The standard error bars for detection of the LFtrnLF are smallest for the NIR 2 means and largest for the NIR 1 means. The standard error bars for recognition of the LFtrnLF are smallest for the FIR means and largest for the NIR 1 means. The mean detection and recognition distances of the LFtrnLF for NIR 1 are significantly higher than the corresponding distances for FIR and HLB. The range of detection distances for the LFtrnRT is from 180 feet for the FIR system to 247 feet for the NIR 1 system. The range of recognition distances for the LFtrnRT is from 144 feet for the FIR system to 225 feet for the NIR 1 system. The standard error bars for detection and recognition of the LFtrnRT are smallest for the HLB means and largest for the NIR 1 means. There are no significantly different distances among any of the VESs for detection and recognition of the LFtrnRT. Back to Figure 17.

Figure 18. Bar graph. Blue pedestrian in right turn: left and right side detection and recognition distances. The graph is titled "Pedestrians in Right Turn (1,250-meter (4,101-foot) radius) Left Side and Right Side." The Y-axis indicates mean distance in feet. The X-axis indicates the four vision enhancement systems evaluated in the study. The left half of the graph presents data for detection and recognition of the RTtrnLF (right turn left) object (there are two bars for each VES, one for detection and one for recognition), and the right half of the graph presents data for detection and recognition of the RTtrnRT (right turn right) object. Each vision enhancement system is represented by two bars, one for detection and one for recognition. Standard error bars are also provided for each graph bar. Letters above each bar note the groupings for the means (means with the same letter are not significantly different) based on the Student-Newman-Keuls results. Capital letters indicate detection groupings, and lowercase letters indicate recognition groupings. The range of detection distances for the RTtrnLF is from 201 feet for the FIR system to 346 feet for the NIR 1 system. The range of recognition distances for the RTtrnLF is from 180 feet for the FIR system to 306 feet for the NIR 1 system. The standard error bars for detection and recognition of the RTtrnLF are smallest for the HLB means and largest for the NIR 2 means. The mean detection distance of the RTtrnLF for NIR 1 is significantly higher than the corresponding distances for FIR and HLB. There are no significantly different distances between any of the VESs for recognition of the RTtrnLF. The range of detection distances for the RTtrnRT is from 137 feet for the NIR 2 system to 243 feet for the NIR 1 system. The range of recognition distances for the RTtrnRT is from 135 feet for the NIR 2 system to 220 feet for the NIR 1 system. The standard error bars for detection and recognition of the RTtrnRT are smallest for the HLB means and largest for the FIR means. There are no significantly different distances among any of the vision enhancement systems for detection and recognition of the RTtrnRT. Back to Figure 18.

Figure 19. Bar graph. Mean subjective ratings by VES for statement 1: "This vision enhancement system allowed me to detect objects sooner than my regular headlights." On the graph, the Y-axis indicates the four VESs evaluated in the study. The X-axis shows the average (mean) rating for the statement. The ratings are on a linear Likert-type scale starting on the left at 1 (Strongly Agree) and ending on the right at 7 (Strongly Disagree). A lower rating indicates a more desirable system. Standard error bars are also provided for each graph bar. Letters to the right of each bar note the groupings for the means (means with the same letter are not significantly different) based on the Student-Newman-Keuls results. The range of mean ratings is from 1.5 for the NIR 1 system to 2.9 for the HLB system. The standard error bar is smallest for the NIR 1 mean and largest for the FIR. The mean rating for NIR 1 is significantly lower than the mean ratings for all other vision enhancement systems. There are no significant differences among the mean ratings for the remaining three vision enhancement systems. Back to Figure 19.

Figure 20. Bar graph. Mean subjective ratings by vision enhancement system for statement 2: "This vision enhancement system allowed me to identify objects sooner than my regular headlights." On the graph, the Y-axis indicates the four vision enhancement systems evaluated in the study. The X-axis shows the average (mean) rating for the statement. The ratings are on a linear Likert-type scale starting on the left at 1 (Strongly Agree) and ending on the right at 7 (Strongly Disagree). A lower rating indicates a more desirable system. Standard error bars are also provided for each graph bar. Letters to the right of each bar note the groupings for the means (means with the same letter are not significantly different) based on the Student-Newman-Keuls results. The range of mean ratings is from 1.5 for the NIR 1 system to 3.0 for the FIR system. The standard error bar is smallest for the NIR 1 mean and largest for the FIR. The mean rating for NIR 1 is significantly lower than the mean ratings for all other vision enhancement systems. There are no significant differences among the mean ratings for the remaining three vision enhancement systems. Back to Figure 20.

Figure 21. Bar graph. Mean subjective ratings by vision enhancement systems for statement 5: "This vision enhancement system did not cause me any more visual discomfort than my regular headlights." On the graph, the Y-axis indicates the four vision enhancement systems evaluated in the study. The X-axis shows the average (mean) rating for the statement. The ratings are on a linear Likert-type scale starting on the left at 1 (Strongly Agree) and ending on the right at 7 (Strongly Disagree). A lower rating indicates a more desirable system. Standard error bars are also provided for each graph bar. Letters to the right of each bar note the groupings for the means (means with the same letter are not significantly different) based on the Student-Newman-Keuls results. The range of mean ratings is from 1.7 for the HLB system to 2.7 for the FIR and NIR 1 systems. The standard error bar is smallest for the HLB mean and largest for the NIR 1. There are no significant differences among the mean ratings for the four vision enhancement systems.
Back to Figure 21.

Figure 22. Equation. Braking distance approximation. In the equation, d subscript BD equals V squared divided by the following: the sum of f and uppercase G, end of sum, that sum multiplied by lowercase g, multiplied by 2. Back to Figure 22.

Figure 23. Diagram. Graphics for detection distances. This diagram from an overhead perspective illustrates a roadway with various icons representing the objects used in the study. Two partially overlapping sections of road are depicted, one curving left and one going straight. On the straight section of road are icons representing left and right static pedestrians, dynamic pedestrian, and tire tread. On the section of road curving to the left are icons representing left and right pedestrians along a 1,250-meter radius curve. Back to Figure 23.

Figure 24. Diagram. FIR mean detection distances. This diagram from an overhead perspective illustrates a vehicle on three possible roadway configurations, one curving left, one going straight, and one curving right. A measurement grid overlays the diagram. The left side of the grid indicates the mean distances of detection using the FIR system. The right side of the grid presents labels at the approximate stopping distances necessary for various speeds. The icons used to depict each object used in the study are pictured on the road at their appropriate locations along the grid. The tire tread is at a detection distance just under 100 feet (requiring speeds less than about 20 miles per hour to accommodate necessary stopping distances). All other icons (blue left, blue right, left turn left, left turn right, right turn left, right turn right, and dynamic pedestrians) are at detection distances between 100 and 200 feet (requiring speeds less than about 20 to 30 miles per hour to accommodate necessary stopping distances). Back to Figure 24.

Figure 25. Diagram. NIR 1 mean detection distances. This diagram from an overhead perspective illustrates a vehicle on three possible roadway configurations, one curving left, one going straight, and one curving right. A measurement grid overlays the diagram. The left side of the grid indicates the mean distances of detection using the NIR 1. The right side of the grid presents labels at the approximate stopping distances necessary for various speeds. The icons used to depict each object used in the study are pictured on the road at their appropriate locations along the grid. The tire tread is at a detection distance just under 100 feet (requiring speeds less than about 20 miles per hour to accommodate necessary stopping distances). The blue left, blue right, left turn left, left turn right, right turn right, and dynamic pedestrians are at detection distances between 200 and 300 feet (requiring speeds less than about 30 to 40 miles per hour to accommodate necessary stopping distances). The right turn left pedestrian is at a detection distance of about 350 feet (requiring speeds less than 45 miles per hour to accommodate necessary stopping distances). Back to Figure 25.

Figure 26. Diagram. NIR 2 mean detection distances. This diagram from an overhead perspective illustrates a vehicle on three possible roadway configurations, one curving left, one going straight, and one curving right. A measurement grid overlays the diagram. The left side of the grid indicates the mean distances of detection using the NIR 2. The right side of the grid presents labels at the approximate stopping distances necessary for various speeds. The icons used to depict each object used in the study are pictured on the road at their appropriate locations along the grid. The tire tread is at a detection distance just under 100 feet (requiring speeds less than about 20 miles per hour to accommodate necessary stopping distances). The right turn right pedestrian is at a detection distance of about 150 feet (requiring speeds less than about 25 miles per hour to accommodate necessary stopping distances). The blue left, blue right, left turn left, left turn right, right turn left, and dynamic pedestrians are at detection distances between 200 and 300 feet (requiring speeds less than about 30 to 40 miles per hour to accommodate necessary stopping distances). Back to Figure 26.

Figure 27. Diagram. HLB mean detection distances. This diagram from an overhead perspective illustrates a vehicle on three possible roadway configurations, one curving left, one going straight, and one curving right. A measurement grid overlays the diagram. The left side of the grid indicates the mean distances of detection using the HLB. The right side of the grid presents labels at the approximate stopping distances necessary for various speeds. The icons used to depict each object used in the study are pictured on the road at their appropriate locations along the grid. The tire tread is at a detection distance just under 100 feet (requiring speeds less than about 20 miles per hour to accommodate necessary stopping distances). The pedestrians labeled left turn left, blue left, dynamic, left turn right, blue right, and right turn right are at detection distances between 100 and 200 feet (requiring speeds less than about 20 to 30 miles per hour to accommodate necessary stopping distances). The right turn left pedestrian is at a detection distance of between 200 and 250 feet (requiring speeds less than 35 miles per hour to accommodate necessary stopping distances). Back to Figure 27.

Appendix C. Contrast Sensitivity Test Diagram. The diagram shows the form the experimenters used to document the results for each participant’s contrast sensitivity exam. The diagram shows a graph with contrast sensitivity from 3 to 300 on the left Y-axis, contrast threshold from .3 to .003 on the right Y-axis, and spatial frequency (cycles per degree) from .5 to 6 on the X-axis. There is a column of numbered circles above each cycle per degree, indicating participant response. There are two diagrams, one for recording the results of the right eye and one for the results of the left eye. Back to Diagram.

ResearchFHWA
FHWA
United States Department of Transportation - Federal Highway Administration