In: Wagner W., Székely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Voi. XXXVIII, Part 7B 
469 
REAL-TIME IMAGE PROCESSING FOR ROAD TRAFFIC DATA EXTRACTION FROM 
AERIAL IMAGES 
D. Rosenbaum, J. Leitloff, F. Kurz, O. Meynberg, and T. Reize 
DLR - German Aerospace Center, Remote Sensing Technology Institute, 
Münchner Str. 20, 82234, Weßling, Germany 
Commission VII Symposium 2010 
KEY WORDS: Monitoring, Recognition, Orthorectification, Georeferencing, Image, Pattern, Sequences, Tracking 
ABSTRACT: 
A world with growing individual traffic requires sufficient solutions for traffic monitoring and guidance. The actual ground based 
approaches for traffic data collection may be barely sufficient for everyday life, but they will fail in case of disasters and mass events. 
Therefore, a road traffic monitoring solution based on an airborne wide area camera system has been currently developed by DLR. 
Here, we present a new image processing chain for real-time traffic data extraction from high resolution aerial image sequences with 
automatic methods. This processing chain is applied in a computer network as part of an operational sensor system for traffic monitoring 
onboard a DLR aircraft. It is capable of processing aerial images obtained with a frame rate of up to 3 Hz. The footprint area of the 
three viewing directions of an image exposure with three cameras is 4 x 1 km at a resolution of 20 cm (recorded at a flight height 
of 1500 m). The processing chain consists of a module for data readout from the cameras and for the synchronization of the images 
with the GPS/IMU navigation data (used for direct georeferencing) and a module for orthorectification of the images. Traffic data 
is extracted by a further module based on a priori knowledge from a road database of the approximate location of road axes in the 
georeferenced and orthorectified images. Vehicle detection is performed by a combination of Adaboost using Haar-like features for 
pixel wise classification and subsequent clustering by Support Vector Machine based on a set of statistical features of the classified 
pixel. In order to obtain velocities, vehicle tracking is applied to consecutive images after performing vehicle detection on the first 
image of the burst. This is done by template matching along a search space aligned to road axes based on normalized cross correlation 
in RGB color space. With this processing chain we are able to obtain accurate traffic data with completeness and correctness both higher 
than 80 % at high actuality for varying and complex image scenes. The proposed processing chain is evaluated on a huge number of 
images including inner city scenes of Cologne and Munich, demonstrating the robustness of our work in operational use. 
1 INTRODUCTION 
Mass events with a large attendance hold in big cities overload 
road infrastructure at regular intervals, since metropolis roads are 
used to full capacity in normal course of life. In case of disas 
ter, maybe with parts of the road network being impassable, total 
collapse of road traffics menaces. Both cases require sufficient 
methods of traffic monitoring and guidance. Common infrastruc 
ture for traffic monitoring like induction loops and video cameras 
is ground based and mainly distributed on main roads. New mon 
itoring approaches collect data by means of mobile measurement 
units which flow with the traffic as test particles for local traffic 
situations and travel times. The so called floating car data (FCD, 
e.g. Schaefer et al., 2002; Busch et al., 2004) obtained from taxi 
cabs can deliver useful traffic information within cities, but they 
are only available in few big cities today. 
These methods and sensors are suited for everyday life, but traf 
fic monitoring and guidance based on these sensors may fail in 
case of mass events or disaster. In those cases a coverage as com 
plete as possible of road level of service is mandatory. Such a 
complete coverage could not be provided by common sensor net 
works with their low spacial resolution data obtained by punctual 
traffic registration limited to main roads as provided by ground 
based sensor networks. In case of disaster, sensor networks based 
on ground infrastructure may fail completely. Not only publicity 
but especially security authorities and organizations which have 
to coordinate and route action and relief forces into and within 
affected areas require precise traffic information. Furthermore, a 
spacial complete area wide traffic surveillance at high actuality 
provides the possibility to generate precise predictions of traffic 
situation in near future by simulations (e.g. Behrisch et al., 2008). 
For these purposes airborne and satellite based solutions for wide 
area traffic monitoring have been produced or are currently un 
der development at DLR. In Reinartz et al. (2006) the general 
suitability of image time series from airborne cameras for traffic 
monitoring was shown. Tests with several camera systems and 
various airborne platforms, as well as the development of an air 
borne traffic monitoring system and thematic image processing 
software for traffic parameters were performed within the projects 
’’LUMOS” and ’’Eye in the Sky” (Ernst et al., 2003; Borner et al., 
2004). The actual project for airborne traffic monitoring and sim 
ulation is called ’’VABENE” (German: Verkehrsmanagement bei 
GroBereignissen und Katastrophen, that means: traffic manage 
ment under mass event and disaster conditions). Aim of the project 
among others is to develop operational airborne optical and radar 
systems for automatic traffic data extraction in real time. First 
results on traffic monitoring based on remote sensing synthetic 
aperture radar (SAR) systems were already shown in e.g. Bethke 
et al. (2007) or Suchandt et al. (2006). On the optical regime, a 
first proposal for a prototype processing chain capable of traf 
fic data extraction from sequences of optical aerial images in 
near real time was shown in Rosenbaum et al. (2008). There, 
algorithms and methods for edge and line based roadside ex 
traction, vehicle detection based on edge detection and geom 
etry validation, and vehicle tracking by template matching us 
ing a normalized cross correlation operator were presented. That 
template matching for car tracking works sufficiently had already 
been proven in Lenhart et al. (2008). There, a similar algorithm 
using a shape based matching operator was introduced. Fur 
thermore, different methods for validation of potential vehicle 
tracks and vehicle detections for outlier elimination were pre 
sented. Some of them are used in the present processing chain.