Previous [ 1] [ 2] [ 3] [ 4] [ 5] [ 6] [ 7] [ 8] [ 9] [ 10] [ 11] [ 12] [ 13] [ 14] [ 15] [ 16]

Journal of Information Science and Engineering, Vol. 32 No. 4, pp. 1113-1127 (July 2016)

A New Approach for Learning Discriminative Dictionary for Pattern Classification

1Faculty of Information Technology
Vietnam National University of Agriculture
Hanoi, 131100 Vietnam
2School of Information and Communication Technology
Hanoi University of Science and Technology
Hanoi, 100000 Vietnam
E-mail:; {binhht; sangdv}

Dictionary learning (DL) for sparse coding based classification has been widely re-searched in pattern recognition in recent years. Most of the DL approaches focused on the reconstruction performance and the discriminative capability of the learned dictionary. This paper proposes a new method for learning discriminative dictionary for sparse rep-resentation based classification, called Incoherent Fisher Discrimination Dictionary Lear- ning (IFDDL). IFDDL combines the Fisher Discrimination Dictionary Learning (FDDL) method, which learns a structured dictionary where the class labels and the discrimina-tion criterion are exploited, and the Incoherent Dictionary Learning (IDL) method, which learns a dictionary where the mutual incoherence between pairs of atoms is exploited. In the combination, instead of considering the incoherence between atoms in a single shared dictionary as in IDL, we propose to incorporate the incoherence between pairs of atoms within each sub-dictionary, which represent a specific object class. This aims to increase discrimination capacity between basic atoms in sub-dictionaries. The combination allows one to exploit the advantages of both methods and the discrimination capacity of the en-tire dictionary. Extensive experiments have been conducted on benchmark image data sets for Face recognition (ORL database, Extended Yale B database, AR database) and Digit recognition (the USPS database). The experimental results show that our proposed method outperforms most of state-of-the-art methods for sparse coding and DL based classification, meanwhile maintaining similar complexity.

Keywords: dictionary learning, sparse coding, fisher criterion, pattern recognition, object classification

Full Text () Retrieve PDF document (201607_16.pdf)

Received February 15, 2015; revised June 18, 2015; accepted July 9, 2015.
Communicated by Chu-Song Chen.