When a road accident occurs, authorities must close the road to traffic, sometimes for hours at a time, because they need to ensure that safety is restored, but they must also investigate the scene of the accident to determine why and how the accident occurred, which can take hours. This not only affects the people involved in the accident, but also others, as it creates road congestion, diversions, delays, causing much dissatisfaction and loss of working hours.
The first important part of the process is that nothing “disappears” from the scene, so it is necessary to map the scene as quickly as possible to capture everything and allow all details to remain intact.
The second important aspect is to collect as much detail as possible, and the data is then used by many users, including police forces and insurance companies for specific calculations to determine the reasons why an accident occurred.
The third important aspect is that investigators are provided with professional tools to help them process the data collected, i.e. to accurately and correctly measure and calculate the distance between the parties involved in the accident, the angle of impact, the braking marks and the final resting positions. Only through precise measurements can they accurately determine the reasons why an accident occurred, so that calculations can then be made ‘off-site’ in the office.
Project description
Problem
Until now, the local forensic investigation department had two options when investigating a traffic accident:
Folosirea 3d laser scanner, which gives good results but is expensive, so the limited number of devices available causes delays and complex resource planning, transferring devices from one part of the country to another.
Abordarea traditional means hand sketch + about 100 photographs on the spot. This method provides documentation of the road accident site, but does not provide good situational awareness and sketching the larger site can be very time consuming with limited accuracy.
Approach
The local police department wanted to understand how and if photogrammetry can help them with easier and more accurate data collection, calculating crash angles, car speeds, identifying objects and measuring and calculating distances, using only photos from drone, photos from the ground and photos from witnesses. The Police Department wanted to understand how much faster the crash site can be removed and clear traffic and also how easily, quickly and accurately calculations can be made with photogrammetric software.
Mapping the scene
The photogrammetry process starts by importing all the photos taken at the accident scene into the photogrammetric software. The packet fitting step helps recognize and find common points between imported images, and then matches them and generates a raw 3D model called a “point cloud”.
The next step is reconstruction which creates a highly detailed 3D point cloud and generates a 3D model of the scene. This 3D model is a perfect input data type for generating full textured 3D models. In the measurement process, forensic experts use built-in CAD functions to obtain accurate details for accident reconstruction, such as angles, distances and more.
Image processing is fast, on the order of minutes, at the site where the photos were purchased. Off-site, in the office, full data processing takes about 3 hours. Deliverables include a 3D model in OBJ format, a 3D sketch with vector CAD lines containing all details, distances, angles, measurements and relationships between objects, DXF export format, an orthophoto map in JPG and KMZ format ready for Google Earth. These can be used to create accident maps or for distribution to other stakeholders.
Benefits of the photogrammetry process include immediate visualization of the forensic medical scene and use of scene mapping for courtroom presentation. This process also saves time in data collection and processing and provides the best possible accuracy and precision in calculating forces and distance measurements. In addition, acquisition costs are lower compared to other modern solutions.