CONDENSATION—Conditional Density Propagation forVisual Tracking
International Journal of Computer Vision
Spatial Color Indexing and Applications
International Journal of Computer Vision
Contour Tracking by Stochastic Propagation of Conditional Density
ECCV '96 Proceedings of the 4th European Conference on Computer Vision-Volume I - Volume I
IEEE Transactions on Pattern Analysis and Machine Intelligence
Distinctive Image Features from Scale-Invariant Keypoints
International Journal of Computer Vision
Multiple Collaborative Kernel Tracking
CVPR '05 Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05) - Volume 2 - Volume 02
Spatiograms versus Histograms for Region-Based Tracking
CVPR '05 Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05) - Volume 2 - Volume 02
The Pyramid Match Kernel: Discriminative Classification with Sets of Image Features
ICCV '05 Proceedings of the Tenth IEEE International Conference on Computer Vision - Volume 2
Covariance Tracking using Model Update Based on Lie Algebra
CVPR '06 Proceedings of the 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Volume 1
Robust Fragments-based Tracking using the Integral Histogram
CVPR '06 Proceedings of the 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Volume 1
Beyond Bags of Features: Spatial Pyramid Matching for Recognizing Natural Scene Categories
CVPR '06 Proceedings of the 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Volume 2
| Hi-index | 0.00 |
We propose a pyramid based representation together with a partial weighted similarity measure to address the partial occlusion problem in visual tracking. In current histogram based tracking methods, due to the lost of spatial information, occlusion cannot be handled very well. In this paper, the spatial pyramid based histogram representation is introduced into the Bayesian inference framework, in which approximate global spatial information is retained and multi-scale information is explored. Correspondingly, considering that the visible parts of a occluded target with the high resolution should have more contribution to the similarity judgment, a partial weighted similarity measure is proposed, in which more confidence is assigned to the better matched parts under the finer scale. The Bayesian inference framework i.e. particle filter then leads the sample distributions to propagate over time. Experiments in different challenging sequences show that our algorithm is superior to the state of art histogram based methods, especially when occlusion happens and a little appearance change exists.