Researchers at the Remote Sensing Technology Institute (IMF) have successfully used a new real-time camera system for traffic monitoring from a helicopter. High resolution images and video clips are processed within a few seconds on board the helicopter and then transmitted to ground. With such data available, traffic control centres can suitably direct traffic related to major public events or in disaster situations as well as rapidly guide rescue teams to their destinations. The new camera system was installed for the first time on a helicopter, DLR‘s BO 105 research aircraft, as part of the VABENE++ traffic research project.
Determining the flow velocities of people and vehicle traffic
The camera system can record motor vehicles as well as people. This makes it possible to evaluate the density of masses of people as well as how fast a group of people or road traffic is moving. This is important information for response teams dealing with major public events or disasters. Directly on board, the collected image data are analysed and georeferenced, in other words related to a map grid, so they can immediately be further processed in a Geographic Information System (GIS). In order to assure that the large amounts of data can be promptly relayed to the relevant operation centres the researchers use a microwave link with a transmission rate of 11 megabits per second.
High data quality
In order to record within seconds what is happening on the ground, the VABENE++ system has to automatically identify motor vehicles and people as such. "Vehicles and people have a particular texture in aerial images. For example, we can identify people as small dots which cast a particular type of shadow" explains Dominik Rosenbaum of DLR’s Remote Sensing Technology Institute and VABENE++ subproject leader responsible for information extraction from aerial image and radar data. Before each flight, sample data sets relevant for the impending mission and the associated algorithms are uploaded in order to "train" the image analysis program. The researchers are faced with yet another challenge when determining the speed of vehicles or movement within clusters of people: they have to track a particular vehicle or person from one image to the next and at the same time subtract the movement of the helicopter. "This is accomplished by directly georeferencing the images. From one image to the next we can correctly allocate up to 90 percent of the vehicles and people, which gives us a very good sampling for determining the velocity of vehicle traffic or flows of people."
The camera system is equipped with three off-the-shelf cameras. To obtain overviews the cameras have 50 mm lenses capable of a field of vision up to 104 degrees Using lenses with a 100 mm focal length achieves resolutions down to 3.5 centimetres per pixel at a flight altitude of 500 metres. One of the cameras also records 4k videos, a digital high-definition video format equivalent to four times HDTV resolution.
Lightweight, flexible construction
The system is contained in a housing measuring some 90x60x50 centimetres and mounted on the left side of the BO 105 helicopter, about ten centimetres above the skids. Four absorbers inside the housing isolate the camera system from helicopter vibrations. The housing can be attached quickly and flexibly to the exterior of the helicopter. "The advantages of helicopters are that they are very manoeuvrable and can also hover over a traffic intersection or a crowd of people", explains Rosenbaum. "This means that we are developing a system that can also be used on police helicopters." The scientists tested previous models of the camera system on the Dornier Do 228-212 and Cessna 208B Grand Caravan DLR research aircraft to document the extent of flooding on the Danube, Elbe and Saale rivers in June 2013 as well as other events.
In the VABENE++ (Verkehrsmanagement bei Großereignissen und Katastrophen) project, powerful tools are being developed for public agencies, organizations with responsibilities for assuring public safety and transport authorities when dealing with disasters and major public events. The goal is to efficiently manage also under extreme conditions the necessary rescue logistics and traffic flows in the vicinity in order to guide relief teams quickly to their destinations. Scientists from various DLR institutes collaborate in this research project, supported by the DLR Research Flight Facility. The Remote Sensing Technology chair at Munich Technical University (TUM) and the Institute of Geoinformatics and Remote Sensing at Osnabrück University also participate in VABENE++. Research concentrates primarily on simulation and large-scale traffic modelling, aerial monitoring of road traffic, traffic risk assessment, data fusion and management, as well as on the further development of Web technologies in a GIS environment.