Robust visual tracking based on occlusion detection and particle redistribution

Authors:
Shengping Zhang;Junpeng Wu;Yuan Tian;Shaohui Liu;Xin Sun
Affiliations:
Harbin Institute of Technology, Harbin, China;Harbin Engineering University, Harbin, China;Harbin Institute of Technology, Harbin, China;Harbin Institute of Technology, Harbin, China;Harbin Institute of Technology, Harbin, China
Venue:
ICIMCS '10 Proceedings of the Second International Conference on Internet Multimedia Computing and Service
Year:
2010

Citing 8
Cited 0

Color-Based Probabilistic Tracking

ECCV '02 Proceedings of the 7th European Conference on Computer Vision-Part I
Kernel-Based Object Tracking

IEEE Transactions on Pattern Analysis and Machine Intelligence
Efficient Mean-Shift Tracking via a New Similarity Measure

CVPR '05 Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05) - Volume 1 - Volume 01
Fast Multiple Object Tracking via a Hierarchical Particle Filter

ICCV '05 Proceedings of the Tenth IEEE International Conference on Computer Vision (ICCV'05) Volume 1 - Volume 01
Fast object tracking using adaptive block matching

IEEE Transactions on Multimedia
Robust online appearance models for visual tracking

IEEE Transactions on Pattern Analysis and Machine Intelligence
Fast occluded object tracking by a robust appearance filter

IEEE Transactions on Pattern Analysis and Machine Intelligence
Visual tracking and recognition using appearance-adaptive models in particle filters

IEEE Transactions on Image Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

Occlusion inference is an extremely difficult problem in visual tracking, especially when the target is occluded fully. In this paper, we proposed a novel visual tracking algorithm to solve this problem, which is based on occlusion edge detection and particle redistribution. Firstly, it judges whether there is occlusion or not. Then if the target happen to be occluded, the occlusion edge is able to be detected. Finally particles are assigned to redistribute around the occlusion edge. Once the target appears again from occlusion, the tracker could capture the target at the same moment. The algorithm could make tracking appeared objects or part of objects occurring in any frame. Experimental results verified that the proposed method outperforms traditional particle filter tracker.