France. September 1-3, 2010 
In: Paparoditis N., Pierrot-Deseilligny M.. Mallet C.. Tournaire O. (Eds), 1APRS. Vol. XXXVIII. Part ЗА - Saint-Mandé, France. September 1-3. 2010 
39 
field. 1935: Kockelmann, 
ider congested conditions 
iss of vehicle's spacing is 
tides 
(1) 
vehicle distance in meter, 
the model, function g(v) 
.g. g( 100 km/h) = 100 m, 
: is the vehicle class, e.g. 
can be interpreted in the 
formula (1) means for all 
b “safe distance = half 
already mentioned this 
ic. The B value is ranging 
skas 2009). 
ed as 
(2) 
on of vehicle class / and 
:ion in (2) the weighted 
directly related to the 
segment and thus the 
>n can be derived (see 
he proposed method is 
ly introduced way of 
•e valid: the values are 
ipproaching zero due to 
hange image. Of course, 
icle density d c estimated 
used and thus a speed of 
imated. 
uodel is used: that is a 
cular road segment is 
peed estimate. 
iTS 
ie method several flight 
»cperimental wide angle 
ted on a Do-228 aircraft 
)f these experiments is 
lite large, the evaluation 
s and can therefore be 
3.1 DLR 3K camera system 
The 3K camera system (“3 Kopf’ = “3 head’’) consists of three 
non-metric off-the-shelf cameras (Canon EOS IDs Mark II, 16 
MPixel). The cameras are arranged in a mount with one camera 
looking in nadir direction and two in oblique sideward 
direction, which leads to an increased FOV of max 110°/ 31° in 
across track/flight direction. The camera system is coupled to a 
GPS/IMU navigation system, which enables the direct geo- 
referencing of the 3K optical images. Based on the use of 
50 mm Canon lenses, the pixel size at a flight height of 1000 m 
above ground is approximately 15 cm and the image swath is 
2.8 km. The pixel size increases up to 50 cm and the swath 
width up to 8 km for a flight height of 3 km. For more details 
see (Kurz 2007). 
3.2 Test site and data 
The motorway A8 south of Munich is one of the busiest parts of 
the German motorway network with an average traffic volume 
of around 100.000 vehicles per day. The test site was a 16 km 
motorway section between motorway junctions “Hofolding"’ 
and “Weyam". On 2 nd Sep. 2006. heavy traffic was expected at 
this section caused by homebound travellers in the direction of 
Munich. Three 3K data takes were acquired between 14:01 and 
15:11 from 2000m above ground in three over flights. During 
each over flight, 22 image bursts were acquired each containing 
four consecutive images. The time difference within these 
bursts was 0.7 s, so that each car was monitored at least for 
2.1 s. To collect the reference data each lane was manually 
processed, that means all vehicles were detected in the images 
by visual interpretation and their speeds measured. 
3.3 Traffic parameter extraction 
Results of traffic parameter extraction on the test site are shown 
in Figures 3 and 4. In Figure 3 an overview of results are 
presented for the mosaic of five frames: highway section of 
approximately 4 km length. We see that the estimated speed 
profiles coincide quite well with the reference measurements. In 
Figure 4 speed estimation results are presented for each of five 
segments separately for more details. 
Traffic congestion is defined usually using the average speed or 
the traffic density. Unfortunately, there is no unique definition 
because of different types of roads (highway, city streets) and 
moreover it is usually country dependent. Having the average 
vehicle speed for each road segment the congestion detection is 
a trivial task and can be performed by a simple threshold. For 
example, if the congestion is defined for speed up to 50 km/h 
(for highways), then the red coloured areas in Figures 3 and 4 
can be interpreted as congested ones. 
Quantitative evaluation of speed profiles is presented in 
(Palubinskas 2009). 
3.4 Discussions 
The performance of the proposed method is very dependent on 
the good quality of the geo-referencing of overlapping images 
and the quality of the road data base. 
A priori information concerning vehicle and road parameters 
should be adapted very carefully to the regional traffic 
conditions. 
For the accurate vehicle density estimation the time lag between 
the two image acquisitions should be selected according to the 
constraints presented in (Palubinskas 2010). 
Image based methods (microscopic model) perform normally 
better for a higher resolution (less than 30 cm pixel spacing 
(Rosenbaum 2008)), thus the aircraft flight height should be 
low or equivalently one should take into account the reduced 
image swath. It seems that the proposed model based method is 
not very sensitive to the resolution because it is working on the 
macroscopic model level. 
We would like to mention that possible false alarms may occur 
during traffic classification due to object shadows on the road 
(e.g. tree shadows as can be seen in Figure 4(d)). Because of the 
usage of only single images for traffic classification the 
following objects: objects (not vehicles) on or over the road 
(bridges, signs, constructions) or objects nearby the road (tree 
and house shadows on the roads) cannot be surely separated 
from vehicles. Further research is needed to incorporate more 
information (e.g. GIS, radar sensor data) in this level. 
What concents future work further experiments are planned to 
test the approach for off-nadir scenes and in the cities during 
different environmental conditions. 
Another research direction is aiming at deriving other traffic 
parameters such as traffic density and traffic flow. 
4. CONCLUSIONS 
A new traffic classification and model-based congestion 
detection approach for image time series acquired by the 
airborne optical 3K camera system is introduced. It allows us to 
derive one of the main traffic parameters - the average speed - 
and the vehicle density as an intermediate product. Other 
parameters such as the beginning and end of congestion, length 
of congestion and travel times can be derived easily from these 
results. The method is based on the vehicle detection on road 
segments by change detection of two images with a short time 
lag, usage of a priori information and simple traffic models. 
Experimental results show the great potential of the proposed 
method for the extraction of traffic parameters on highways in 
along-track scenes. The estimated speed profiles coincide 
qualitatively and quantitatively quite well with the reference 
measurements. 
5. ACKNOWLEGMENT 
The authors would like to thank our colleagues Franz Kurz, 
Erich Bogner and Rolf Statter for their efforts in planning the 
flight campaign, data acquisition and data processing. Special 
thanks to our trainee Mantas Palubinskas for the 
implementation and validation of the traffic classification. 
6. REFERENCES 
Ernst, I., Sujew, S., Thiessenhusen, K.U., Hetscher, M., 
RaBmann, S. and Ruhe, M., 2003. LUMOS - airborne traffic 
monitoring system. In: Proceedings of 6th IEEE International 
Conference on Intelligent Transportation Systems, Shanghai, 
China, 2003. 
Greensfield, B.D., 1935. A study of traffic capacity. In: 
Highway Research Board Proceedings, vol. 14. 448:477.