Abstract
Canonical correlations analysis (CCA) is often used for feature extraction and dimensionality reduction. However, the image vectorization of CCA breaks the spatial structure of the original image, and the excessive dimensions of vectors often cause the curse of dimensionality problem. In this paper, we propose a novel feature extraction method based on CCA in multi-linear discriminant subspace by encoding each action sample as a high-order tensor. An optimization approach is presented to iteratively learn the discriminant subspace by unfolding the tensor along different tensor modes, which shows that most of the underlying data structure, including the spatio-temporal information, is retained and the curse of dimensionality problem is alleviated by the use of the proposed approach. At the same time, an incremental scheme is developed for multi-linear subspace online learning, which can improve the discriminative capability efficiently and effectively. In addition, the nearest neighbor classifier (NNC) is employed for action classification. Experiments on the Weizmann database show that the proposed method outperforms the state-of-the-art methods in terms of accuracy and time complexity, and it is robust against partial occlusion.
Original language | English |
---|---|
Pages (from-to) | 56-63 |
Number of pages | 8 |
Journal | Neurocomputing |
Volume | 83 |
Issue number | - |
Early online date | 27 Dec 2011 |
DOIs | |
Publication status | Published - 15 Apr 2012 |
Bibliographical note
A paid open access option is available for this journal.Voluntary deposit by author of pre-print allowed on Institutions open scholarly website and pre-print servers
Voluntary deposit by author of authors post-print allowed on institutions open scholarly website including Institutional Repository
Deposit due to Funding Body, Institutional and Governmental mandate only allowed where separate agreement between repository and publisher exists
Set statement to accompany deposit
Published source must be acknowledged
Must link to journal home page or articles' DOI
Publisher's version/PDF cannot be used
Articles in some journals can be made Open Access on payment of additional charge
NIH Authors articles will be submitted to PubMed Central after 12 months
Authors who are required to deposit in subject-based repositories may also use Sponsorship Option
Keywords
- Canonical correlations analysis
- Multi-linear subspace
- Discriminant information
- Incremental learning
- Action recognition