Robust visual tracking based on feature fusion

被引：0

作者：

Li, Lin ^{[1
,2
]}

Yu, Shengsheng ^{[2
]}

Feng, Qing ^{[2
]}

机构：

[1] School of Computing, Wuhan University of Technology

[2] School of Computing, Huazhong University of Science and Technology

来源：

Journal of Computational Information Systems | 2013年 / 9卷 / 16期

关键词：

Feature fusion; Observation model; Particle filter; Robustness; Visual tracking;

D O I：

10.12733/jcisP0716

中图分类号：

学科分类号：

摘要：

To drive computational visual tracking toward more robust outputs, we need a more accurate and adaptive feature representation of target. In this paper, we propose a tracking algorithm using new multi- feature statistical observation model based on the particle filter framework. Four complementary features are described with histograms and fused in a novel way. We demonstrate how to establish observation model and feature-level fusion. The fusion strategy is based on the variability of the observation likelihood. It improves the reliability estimation of each feature, makes the weight distribution of particles more rational and enhances algorithm robustness. An additional contribution is a novel particle resampling method that is able to optimize particle performance. Experimental results show that the proposed algorithm is more robust, especially in a cluttered environment with varying illumination conditions. © 2013 Binary Information Press.

引用

页码：6527 / 6534

页数：7

共 9 条

[1]

Zhang H.Z., Zhang J.H., Object tracking algorithm based on camshift, Computer Engineering and Design, 27, 11, pp. 2012-2014, (2012)

[2]

Pu F.L., Wang R., Object detection and tracking based on optical flow, Journal of Chinese Peoples Public Security University, 60, 2, pp. 58-60, (2009)

[3]

Gustafsson F., Particle filter theory and practice with positioning applications, Aerospace and Electronic Systems Magazine, 25, 7, pp. 53-82, (2010)

[4]

Exner D., Weimar B., Fast and robust camshift tracking, Proc. Computer Vision and Pattern Recognition Workshop'10, pp. 9-16, (2010)

[5]

Beauchemin S.S., Barron J.L., The computation of optical flow, ACM Computing Surveys, 27, 3, pp. 434-467, (1995)

[6]

Tang D., Zhang Y.J., Combining mean-shift and particle filter for object tracking, Proc. 6th International Conference on Image and Graphics, pp. 771-776, (2011)

[7]

Maggio E., Smerladi F., Cavallaro A., Adaptive multifeature tracking in a particle filtering frame-work, IEEE Trans. on Circuits and System for Video Technology, 17, 10, pp. 1348-1359, (2007)

[8]

Gu X., Wang H.T., Fusing multiple features for object tracking based on uncertainty measurement, Acta Automatica Sinica, 37, 5, pp. 550-559, (2011)

[9]

Yuan L.H., Fu L., Yang Y., Analysis of texture feature extracted by gray level co-occurrence matrix, Journal of Computer Applications, 29, 4, pp. 1018-1021, (2009)

← 1 →