Aerial crash scene mapping with iWitness

Background:

In October 2005, the Minnesota State Patrol (MnSP) teamed with DeChant Consulting Services – DCS Inc to conduct a low-altitude aerial photo-grammetric survey of a staged traffic accident scene. The photogrammetric processing was performed using the iWitness software system.

In some MnSP crash scenes, data collection is achieved by aerial imaging from a helicopter. The MnSP ‘paints’ the road scene at the time of the incident and then later flies over the scene to record imagery, which is typically acquired with a moderately-priced, consumer-grade digital camera. In the interests of providing more efficient incident management, the use of aerial images for accident reconstruction is of particular interest to both the MnSP and Minnesota Department of Transportation (MnDOT). Aerial imaging offers the prospect of being able to re-open the roads more quickly and more safely. Image measurement and data processing can be handled with ease via the inexpensive iWitness system for close-range photogrammetry.

The Survey:

The staged accident scene was 700’ (213m) long. MnSP painted simulated road markings (i.e. tire yaw and scuffmarks, and also the vehicle’s final rest position from the crash event). The paint marks spanned portions of the entire 700’ length of roadway.

In order to provide independent verification of the measurement performance obtained with the iWitness photogrammetric measurement of ground features of interest at the accident scene, a Leica total station was also used to map the scene, as well as adjacent fence lines and additional road detail. One aim of the project was to overlay within a CAD system the iWitness measurements from aerial imagery with the corresponding total station measurements and compare the relative accuracy of the two technologies.

The camera used for the aerial imaging was a Nikon D70 with a 50mm lens. This combination was deemed very suitable for use at a flying height of 500’, since it would yield 3D positioning accuracy of ground features to better than 0.1’. The camera was metrically calibrated via an automated calibration process available in iWitness.

As in all such investigative trials, there were lessons learned from performing the test measurements, especially in regard to overcoming unforeseen changes to design procedures. For example, rather than flying at the desired height of 500’ (roughly 150m) above ground level, the helicopter flew at 1000’ to 1500’ (or roughly 300 to 450m). Thus, the scale of the aerial imagery acquired was only half that expected, with there being a commensurate loss in design accuracy in the subsequent iWitness measurements. Nevertheless, 3D measurement accuracy at the 0.2’ (5cm) level was still achievable.

The issue of the smaller than expected imaging scale was compounded by the consequential fact that some of the yellow paint marks used at the accident scene, namely those on dull concrete surfaces, could not be clearly distinguished in the imagery recorded from flying heights of 1000’ to 1500’. It should be noted that the yellow paint worked well on asphalt portions of the road, which accounted for approximately 95% of the surveyed area. However, the 5% area on concrete surfaces incorporated critical evidence that needed to be mapped, and this was carried out with supplementary ground-based images (using the Nikon D70 with a 20mm lens). Both aerial and terrestrial images were easily combined in the same iWitness project. Six aerial images, as well as five ground-level images were used in the iWitness measurement project. The internal accuracy of measurement was 1:8400 of object size, or just under 0.1’.

Measurement Analysis:

In order to compare the results obtained with iWitness versus those from the total station, checks of measured distances were employed. Initially, the scale of the photogrammetric network was set with one distance (of 588.9’), and 12 distance checks were then made over various distances throughout the full length of the test site. The check distances were mostly from the total station, but a few were tape measurements. The overall RMS discrepancy of distance measurements between the two approaches was 5.6” (14cm), which confirmed that low-level aerial photogrammetry could achieve the accuracy levels desired for traffic accident reconstruction. The iWitness measurement data was then exported to the Crash Zone CAD package (www.cadzone.com) and a plan view showing the outcomes of the two independent measurement sets is shown in the figure below.

Conclusion:

The test conducted confirmed that iWitness could be utilized in conjunction with low-level digital aerial imagery to produce 3D measurements for traffic accident reconstruction to an accuracy level of around 0.1’ (about 3cm). The impressive results obtained required use of only an off-the-shelf, consumer-grade digital camera, with attention being paid to selecting an appropriate lens (50mm focal length in this case) and performing a fully metric calibration, again via iWitness. In spite of the photography in this case being imaged at a higher flying height than desired, sufficiently accurate ground measurements were still achieved. This proved that iWitness is a powerful, practical and flexible 3D measurement tool for diagramming accident scenes from aerial imagery.

The diagram below, is the plan-view combined DXF export results of the iWitness photogrammetry measurements compared to the Minnesota State Patrol's EDM total station. GREEN is the iWitness, and RED is the total station.

The below image is an oblique , reverse angle view of the iWitness measured points and lines as they appear in the iWitness 3D Graphics view. The graphics view denotes the measured roadway geometry and tire marked evidence. Note the four terrestrial-based images (left side) mapping the 'ground painted' car's final rest position and other road evidence deemed important in the mapping project.

The last illustration is also an oblique view of the iWitness measured points and lines in the 3D Graphics view. This view denotes the measured light-rays of one of the camera positions as observed from the helicopter at 1,000 feet elevation. The red line is the project scale distance.

The Minnesota State Patrol's helicopter leaving the scene after 10 minutes of digital imaging by 3 different cameras.

Special thanks to the MnSP Accident Reconstruction Team, and the MnSP Air Wing Unit for supporting this trial scene exercise.

Back to top