me XXXIX-B3, 2012 International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B3, 2012
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia
24
= T ; I I T T
E T femeto atreodotooto-o cashoshodoo j
8 9 10 11 12 13 16
X [m]
= 19 ES EE TES T T du T
E HE A A T j
8 9 10 11 12 13 14 15 16
X [m]
Figure 6: Results for a person approaching the cameras. Reference data is shown in red, our results in blue.
The presented work will be the basis of further investigations on
the directions of the axes. the optimisation of matching costs and the analysis of the derived
. point cloud.
s the quality of the result- ACKNOWLEDGEMENTS
isual inspection. Figure 4
n maximum and minimum This research was funded by the German Federal Ministry of Ed-
f points is 1890 and 5936 ucation and Research (BMBF), 13N10809 - 13N10814.
1d still consists of enough
f the person's position the
' from the heavily reduced REFERENCES
Haritaoglu, I., Harwood, D. and Davis, L., 1998. W*5S: A real-
time system for detecting and tracking people in 2 1/2D. In:
set of the same scene was ECCV’98, LNCS, Vol. 1406, pp. 877-892.
en wider stereo baselines. Hirasaki, E., Moore, S. T., Raphan, T. and Cohen, B., 1999. Ef-
reduced scale and resolu- fects of walking velocity on vertical head and body movements
it clouds are significantly during locomotion. Experimental Brain Research 127, pp. 117-
tructed on a person at an 130.
liable height estimation is
ng results are retrieved for Hirschmiiller, H., 2008. Stereo processing by semiglobal match-
ing and mutual information. IEEE Transactions on Pattern Anal-
ysis and Machine Intelligence 30(2), pp. 328 —341.
Jánen, U., Huy, M., Grenz, C., Háhner, J. and Hoffmann, M.,
2011. Distributed three-dimensional camera alignment in highly-
UTURE WORK dynamical prioritized observation areas. In: Fifth ACM/IEEE In-
ternational Conference on Distributed Smart Cameras (ICDSC).
principal applicability of
à set-ups.
veillance videos produces KaewTraKulPong, P. and Bowden, R., 2001. An improved
cular analysis. Estimates adaptive background mixture model for real-time tracking with
of their body height can shadow detection. In: Proc. 2nd European Workshop on Ad-
tructed on the visible sur- vanced Video Based Surveillance Systems.
from low resolution im- Konolige, K., 1997. Small vision systems: Hardware and imple-
resolution material would mentation. In: Proc. of 8th International Symposium on Robotics
of people in terms of an Research.
re sophisticated approach
seometric primitive to the McGlone, J. C. (ed.), 2004. Manual of Photogrammetry. 5 edn,
American Society for Photogrammetry and Remote Sensing.
Monari, E., Maerker, J. and Kroschel, K., 2009. A robust and
lysis can be tackled with efficient approach for human tracking in multi-camera systems.
reo vision is resource in- In: Proc. of Sixth IEEE International Conference on Advanced
alient people in this work. Video and Signal Based Surveillance, 2009. AVSS '09, pp. 134
] load could be achieved 7139.
xpectable disparity range Zhao, T., Aggarwal, M., Kumar, R. and Sawhney, H., 2005. Real-
ding epochs. time wide area multi-camera stereo tracking. In: IEEE Computer
Society Conference on Computer Vision and Pattern Recogni-
support subsequent monoc- tion, 2005, Vol. 1, pp. 976—983.
| the ground plane can be
f its head. This is particu-
on after the person left the
Zhou, J., Wan, D. and Wu, Y., 2010. The chameleon-like vision
system. IEEE Signal Processing Magazine 27(5), pp. 91 —101.
5i