Searching for just a few words should be enough to get started. If you need to make more complex queries, use the tips below to guide you.
Article type: Research Article
Authors: Liu, Jun-Chia; b; * | Jiao, Li-Chenga | Kang, Jun-Ruib
Affiliations: [a] Key Laboratory of Intelligent Perception and Image Understanding of Ministry of Education, International Research Center for Intelligent Perception and Computation, Joint International Research Laboratory of Intelligent Perception and Computation, Xidian University, Xi’an 710071, Shaanxi, China | [b] Xi’an Modern Control Technology Research Institute, Xi’an 710065, Shaanxi, China
Correspondence: [*] Corresponding author: Jun-Chi Liu, Key Laboratory of Intelligent Perception and Image Understanding of Ministry of Education, International Research Center for Intelligent Perception and Computation, Joint International Research Laboratory of Intelligent Perception and Computation, Xidian University, Xi’an 710071, Shaanxi, China. Tel.: +86 13519104256; E-mail: junchi053@163.com.
Abstract: Discriminative tracking sees object tracking as a binary classification problem and uses a discriminative classifier to separate the target from surrounding background, then uses the positive and negative sample increments extracted by the current frame to update classifier and object location. However, this kind of method exists sample ambiguity. MIL (Multiple Instance Learning) utilizes bag to encapsulate multi-instance and bag tag to replace instance tag, which can be better to solve the ambiguity problem. But when MIL maximizes log value of bag likelihood to select weak classifier, it cannot fully excavate efficient information of feature and the selected feature may not be optimal, which causes error accumulation and final track drifting. The method selection feature based on minimizing trace is proposed in this paper. And information matrix is used to measure the uncertainty of classification model, which not only reduces weak classifier quantity for composing strong classifier, but also ensure that the selected feature has stronger discriminability. It ensures the tracking precision and can efficiently reduce the computational complexity simultaneously. Through the experimental comparisons with some popular tracking algorithms for multiple image sequences including various challenging factors, the superior properties of the algorithm in this paper is verified in the aspects of tracking precision and operating speed.
Keywords: Object tracking, multi-instance learning, minimizing trace, feature selection
DOI: 10.3233/JCM-170739
Journal: Journal of Computational Methods in Sciences and Engineering, vol. 17, no. 3, pp. 519-531, 2017
IOS Press, Inc.
6751 Tepper Drive
Clifton, VA 20124
USA
Tel: +1 703 830 6300
Fax: +1 703 830 2300
sales@iospress.com
For editorial issues, like the status of your submitted paper or proposals, write to editorial@iospress.nl
IOS Press
Nieuwe Hemweg 6B
1013 BG Amsterdam
The Netherlands
Tel: +31 20 688 3355
Fax: +31 20 687 0091
info@iospress.nl
For editorial issues, permissions, book requests, submissions and proceedings, contact the Amsterdam office info@iospress.nl
Inspirees International (China Office)
Ciyunsi Beili 207(CapitaLand), Bld 1, 7-901
100025, Beijing
China
Free service line: 400 661 8717
Fax: +86 10 8446 7947
china@iospress.cn
For editorial issues, like the status of your submitted paper or proposals, write to editorial@iospress.nl
如果您在出版方面需要帮助或有任何建, 件至: editorial@iospress.nl