Information Technology Reference
In-Depth Information
the model, and their segments are used to train the GNG. Opposite to other
trajectory clustering methods described in the literature, we do not need to select
only complete trajectories and the information of these pieces of trajectories are
exploited but as these incomplete trajectories are not frequently repeated they
do not build a prototype.
4 Conclusions
In this paper we have introduced a new method to obtain prototype trajectories.
Based on recent works that suggest that the problem of trajectory matching can
be better solved if pieces of trajectories are considered, we start the problem
by dividing observed trajectories into linear segments. The GNG algorithm is
used then to find sub-trajectory prototypes and demonstrates that has a good
behaviour due to its robustness against outliers and that a number of clusters
is not needed a priori. These prototypes are used to build sequences of common
subtrajectories, very useful to detect events in higher levels. Very promising
results have been obtained using synthetic observed trajectories in real scenes.
Future work will be focused on working with real observed trajectories and
feed back the tracking algorithm with the information provided by the trajectory
identification algorithm, once that the prototypes have been learned.
Acknowledgement
This work was supported by the Spanish Ministry for Science and Innovation
under Projects TIN2010-20845-C03-03, TIN2010-20845-C03-01, Plan Avanza
TSI-020302-2009-59 and by the Madrid Government under project CCG10-
UAH/TIC-5965.
References
1. Antonini, G., Thiran, J.: Counting pedestrians in video sequences using trajectory
clustering. IEEE Transactions on Circuits and Systems for Video Technology 16(8),
1008-1020 (2006)
2. Atev, S., Masoud, O., Papanikolopoulos, N.: Learning trac patterns at inter-
sections by spectral clustering of motion trajectories. In: IEEE/RSJ International
Conference on Intelligent Robots and Systems, pp. 4851-4856 (2006)
3. Atev, S., Miller, G., Papanikolopoulos, N.: Clustering of vehicle trajectories. IEEE
Transactions on Intelligent Transportation Systems 11(3), 647-657 (2010)
4. Bashir, F.I., Khokhar, A., Schonfeld, D.: Object trajectory-based activity classifi-
cation and recognition using hidden markov models. IEEE Transactions on Image
Processing 16(7), 1912-1919 (2007)
5. Buzan, D., Sclaroff, S., Kollios, G.: Extraction and clustering of motion trajectories
in video. In: International Conference on Pattern Recognition, pp. 521-524 (2004)
6. Canales, F., Chacon, M.: Modification of the growing neural gas algorithm for
cluster analysis. In: Rueda, L., Mery, D., Kittler, J. (eds.) CIARP 2007. LNCS,
vol. 4756, pp. 684-693. Springer, Heidelberg (2007)
Search WWH ::
Custom Search