Abstract
We suggest a method handling occlusion in visual tracking. The occlu-
sion handling method considers the joint appearance of overlapping objects
to estimate the position of the target. That is, we can track the object effec-
tively even it is occluded by other objects. For two objects, the appearance
of the overlapping area of two objects belongs to one of them. We use the
corresponding joint appearance to compute the likelihood in these two cases.
We select the maximum of two likelihoods as the winner to decide which is
occluding the other object. Besides, joint appearance model provides the
information about the depth ordering to help meaningfully setting the pa-
rameters for occluded or occluding cases. We also talk about the case, when
the object is occluded by background regions. We use robust statistics to
maintain the performance. We combine joint appearance model and robust
statistics in a unified framework to solve the occlusion problem.

Abstract
We suggest a method handling occlusion in visual tracking. The occlu-
sion handling method considers the joint appearance of overlapping objects
to estimate the position of the target. That is, we can track the object effec-
tively even it is occluded by other objects. For two objects, the appearance
of the overlapping area of two objects belongs to one of them. We use the
corresponding joint appearance to compute the likelihood in these two cases.
We select the maximum of two likelihoods as the winner to decide which is
occluding the other object. Besides, joint appearance model provides the
information about the depth ordering to help meaningfully setting the pa-
rameters for occluded or occluding cases. We also talk about the case, when
the object is occluded by background regions. We use robust statistics to
maintain the performance. We combine joint appearance model and robust
statistics in a unified framework to solve the occlusion problem.

Bibliography
[1] Shai Avidan. Support vector tracking. IEEE Transactions on Pattern
Analysis and Machine Intelligence, 26(8):pp. 1064–1072, 2001.
[2] Andrew Blake and Michael Isard. Active Contours. Springer-Verlag,
1998.
[3] Vicent Caselles, Ron Kimmel, and Guillermo Sapiro. Geodesic active
contours. International Journal of Computer Vision, 22(1):pp. 61–79,
1997.
[4] Robert T. Collins, Alan J. Lipton, Takeo Kanade, Hironobu Fujiyoshi,
David Duggins, Yanghai Tsin, David Tolliver, Nobuyoshi Enomoto, Os-
amu Hasegawa, Peter Burt, and Lambert Wixson. A system for video
surveillance and monitoring. Carnegie Mellon University, Pittsburgh,
PA, Tech. Rep., CMU-RI-TR-00-12, 2000.
[5] J M Ferryman, S Maybank, and A Worrall. Visual surveillance for
moving vehicles. International Journal of Computer Vision, 37(2):pp.
187–197, 2000.
[6] Christophe Fiorio and Jens Gustedt. Two linear time union-find strate-
gies for image processing. Theoretical Computer Science, 154(2):pp. 165
– 181, 1996.
[7] David A. Forsyth and Jean Ponce. Computer Vision: A Modern Ap-
proach. Prentice Hall, 2002.
[8] Gregory D. Hager and Peter N. Belhumeur. Efficient region tracking
with parametric models of geometry and illumination. IEEE Transac-
tions on Pattern Analysis and Machine Intelligence, 20(10):pp. 1025–
1039, 1998.
[9] Weiming Hu, Tieniu Tan, LiangWang, and Steve Maybank. A survey on
visual surveillance of object motion and behaviors. IEEE Transactions
on Systems, Man, and Cybernetics, 34(3):pp. 334–352, 2004.
[10] Peter J. Huber. Robust Statistics. Wiley, 1981.
[11] Michael Isard and Andrew Blake. CONDENSATION—conditional den-
sity propagation for visual tracking. International Journal of Computer
Vision, 29(1):pp. 107–112, 1998.
[12] Allan D. Jepson, David J. Fleet, and Michael J. Black. A layered motion
representation with occlusion and compact spatial support. European
Conference on Computer Vision, 1:pp. 692–706, 2002.
[13] Allan D. Jepson, David J. Fleet, and Thomas F. El-Maraghi. Robust
online appearance models for visual tracking. IEEE Conference on Com-
puter Vision and Pattern Recognition, 1:pp. 415–422, 2001.
[14] Rudolph Emil Kalman. A new approach to linear filtering and prediction
problems. Transactions of the ASME–Journal of Basic Engineering,
82(Series D):pp. 35–45, 1960.
[15] Jongwoo Lim, David Ross, Ruei Sung Lin, and Ming Hsuan Yang. In-
cremental learning for visual tracking. Advances in Neural Information
Processing Systems, 17:pp. 793–800, 2005.
[16] D. Meyer, J. Denzier, and H. Niemann. Model based extraction of
articulated objects in image sequences for gait analysis. In Proc. IEEE
International Conference Image Processing, pages pp. 78–81, 1998.
[17] Christopher Rasmussen and Gregory D. Hager. Probabilistic data asso-
ciation methods for tracking complex visual objects. IEEE Transaction
on Pattern Analysis and Machine Intelligence, 23(6):pp. 560–576, 2001.
[18] C. Stauffer and W. Grimson. Adaptive background mixture models for
real-time tracking. In Proc. IEEE Conf. Computer Vision and Pattern
Recognition, 2:pp. 246–252, 1999.
[19] Kenji Suzuki, Isao Horiba, and Noboru Sugie. Linear-time connected-
component labeling based on sequential local operations. Computer Vi-
sion and Image Understanding, 89:pp. 1–23, 2003.
[20] Hai Tao, Harpreet S. Sawhney, and Rakesh Kumar. Object tracking with
bayesian estimation of dynamic layer representations. IEEE Transac-
tions on Pattern Analysis and Machine Intelligence, 24(1):pp. 75–89,
2002.
[21] John Y. A. Wang and Edward H. Adelson. Representing moving im-
ages with layers. IEEE Transactions on Image Processing Special Issue:
Image Sequence Compression, 3(5):pp. 625–638, 1994.
[22] Greg Welch and Gary Bishop. An introduction to the kalman filter.
2004.
[23] Oliver Williams, Andrew Blake, and Roberto Cipolla. A sparse prob-
abilistic learning algorithm for real-time tracking. IEEE International
Conference on Computer Vision, 1:pp. 353–360, 2003.
[24] Shaohua Zhou, Rama Chellappa, and Baback Moghaddam. Visual track-
ing and recognition using appearance-adaptive models in particle filters.
IEEE Transaction on Image Processing, 13(11):pp. 1491–1506, 2004.
[25] Yue Zhou and Hai Tao. A background layer model for object tracking
through occlusion. In Proceedings of IEEE International Conference on
Computer Vision, pages pp. 1079–1085, 2003.