Class relatedness oriented-discriminative dictionary learning for multiclass image classification. (November 2016)

• Record Type: Journal Article
• Authors: Zhang, Dongyu; Liu, Pengju; Zhang, Kai; Zhang, Hongzhi; Wang, Qing; Jing, Xiaoyuan
• Is Part Of: Pattern recognition. Volume 59(2016:Nov.)
• Page Start: 168
• Page End: 175
• Publication Date: 2016-11
• Subjects: Dictionary learning -- Joint sparsity -- ℓ1, ∞-norm -- Support vector machine -- Class relatedness; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2015.12.005
• Languages: English
• ISSNs: 0031-3203

Abstract

Abstract: Dictionary learning (DL) has recently attracted intensive attention due to its representative and discriminative power in various classification tasks. Although much progress has been reported in the existing supervised DL approaches, it is still an open problem that how to build the relationship between dictionary atoms and the class labels in multiclass classification. In this paper, based on the assumption that the relevance of dictionary atoms could be helpful in multiclass classification task, we proposed a class relatedness oriented (CRO) discriminative dictionary learning method for sparse coding. Utilizing the ℓ 1, ∞ -norm regularization on the coding coefficient matrix, the proposed method can adaptively learn the class relatedness between dictionary atoms and the multiclass labels. Experimental results of face recognition, object classification, and action recognition demonstrate that our proposed method is comparable to many state-of-the-art DDL methods. Abstract : Highlights: We propose a new discriminative dictionary learning method for image classification. We show that the proposed method can adaptively learn the relatedness between dictionary atoms and the class labels. The proposed method achieves state-of-the-art performance on several public databases.