Motion segmentation using GPCA techniques and optical flow

ABSTRACT

In this work, the use of the Generalized Principal Components Analysis (G-PCA) to improve the segmentation of moving objects in image sequences is proposed. In order to obtain this improvement, the noise components in the image derivatives are reduced, and afterwards, a method based on linear algebra is used to make the segmentation. Furthermore this work presents diverse tests to compare the results reached with and without the noise reduction in the image derivatives, and using the nonlinear minimization of an error function. A remarkable improvement in the segmentation quality and the processing time can be observed in every experiment when using the proposed method.

REFERENCES

• 1: Black, M. and Anandan, P. "Robust dynamic motion estimation over time". Proceedings of the IEEE International Conference on Computer Vision and Pattern Recognition. June 1991. Page(s): 296--302.
• 2: Yan, H.; Tjahjadi, T.; "Multiple motion segmentation through a highly robust estimator". IEEE International Conference on Systems, Man and Cybernetics Oct. 2004. Page(s): 3082--3087.
• 3: Darrell, T.; Pentland, A. "Robust estimation of a multi-layered motion representation". Proc. of the IEEE Workshop on Visual Motion. 1991. Page(s): 173--178.
• 4: Yair Weiss , Edward H. Adelson, A Unified Mixture Framework for Motion Segmentation: Incorporating Spatial Coherence and Estimating the Number of Models, Proceedings of the 1996 Conference on Computer Vision and Pattern Recognition (CVPR '96), p.321, June 18-20, 1996
• 5: Boult, T. E. and Brown, L. G. "Factorization-based segmentation of motions". Proc. of the IEEE Workshop on Motion Understanding. pages 179--186 (1991)
• 6: João Paulo Costeira , Takeo Kanade, A Multibody Factorization Method for Independently Moving Objects, International Journal of Computer Vision, v.29 n.3, p.159-179, Sept. 1998 [doi>10.1023/A:1008000628999]
• 7: Kanatani, K. "Motion Segmentation by Subspace Separation and Model Selection". Proceedings of the 8th IEEE International Conference on Computer Vision 2001. Volume 2, Page(s): 586--591
• 8: René Vidal , Shankar Sastry, Segmentation of Dynamic Scenes from Image Intensities, Proceedings of the Workshop on Motion and Video Computing, p.44, December 05-06, 2002
• 9: Rene Vidal , Yi Ma , Shankar Sastry, Generalized Principal Component Analysis (GPCA), IEEE Transactions on Pattern Analysis and Machine Intelligence, v.27 n.12, p.1945-1959, December 2005 [doi>10.1109/TPAMI.2005.244]
• 10: Matthew Turk , Alex Pentland, Eigenfaces for recognition, Journal of Cognitive Neuroscience, v.3 n.1, p.71-86, Winter 1991 [doi>10.1162/jocn.1991.3.1.71]
• 11: Jieping Ye , Ravi Janardan , Qi Li, GPCA: an efficient dimension reduction scheme for image compression and retrieval, Proceedings of the tenth ACM SIGKDD international conference on Knowledge discovery and data mining, August 22-25, 2004, Seattle, WA, USA [doi>10.1145/1014052.1014092]

INDEX TERMS

Keywords:
GPCA, motion segmentation, optical flow