Robust visual tracking combining global and local appearance models

Authors:
Zhisheng Xu;Shengping Zhang;Julong Pan;Xin Sun;Shaohui Liu
Affiliations:
Harbin Institute of Technology, Harbin, China;Harbin Institute of Technology, Harbin, China;Zhejiang University, Hangzhou, China and China Jiliang University, Hangzhou, 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 5
Cited 0

Histograms of Oriented Gradients for Human Detection

CVPR '05 Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05) - Volume 1 - Volume 01
Decentralized Multiple Target Tracking Using Netted Collaborative Autonomous Trackers

CVPR '05 Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05) - Volume 1 - Volume 01
MCMC-Based Particle Filtering for Tracking a Variable Number of Interacting Targets

IEEE Transactions on Pattern Analysis and Machine Intelligence
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
A tutorial on particle filters for online nonlinear/non-GaussianBayesian tracking

IEEE Transactions on Signal Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, we present a robust visual tracking method combining global and local appearance models. We model the object to be tracked with a RGB color histogram and multiple histograms of oriented gradients (HOG). Modeling object using only the former, a global appearance model, is widely used in visual tracking. However, it suffers many challenges such as illumination changes and pose changes and so on. In order to overcome this problem, we also model the object with multiple block based HOG histograms. The HOG histogram is a local appearance model and can effectively represent the shape information of the object which also gain increasing interests in computer vision especially in pedestrian detection. These two appearance models are complementary and used in the particle filter tracking framework. We test the performance of the proposed method on several challenging sequences, which verifies that our method outperforms the standard particle filter and achieves significant improvement.