Unified discriminating feature analysis for visual category recognition

Wenhe Liu, Chenqiang Gao, Xiaojun Chang, Qun Wu

Research output: Contribution to journalArticleResearchpeer-review

5 Citations (Scopus)


Visual category recognition (VCR) is one of the most important tasks in image and video indexing. To deal with high dimension image/video data, feature analysis algorithms have been widely used for visual category recognition. In this paper, to enhance the flexibility regarding the exploitation of labeled or unlabeled data, we propose a unified feature analysis framework that can be applied to both supervised and semi-supervised scenarios. Furthermore, by revealing intrinsic relationships of traditional feature analysis methods, our framework not only integrates traditional methods, but also introduces an ℓ2,1-norm regularization term for sparse learning. Extensive experiments report that the proposed method obtains advantageous performance in comparison with other state-of-the-art supervised and semi-supervised feature selection algorithms.

Original languageEnglish
Pages (from-to)772-778
Number of pages7
JournalJournal of Visual Communication and Image Representation
Issue numberPart B
Publication statusPublished - Oct 2016
Externally publishedYes


  • Feature selection
  • Semi-supervised learning
  • Sparse learning
  • Supervised learning

Cite this