The sparse representation based classification (SRC) method and collaborative representation based classification (CRC) method attract more and more attention in recent years, due to their promising result and robustness for face recognition. However, both SRC and CRC algorithms directly use the training samples as the dictionary, which leads to large fitting error. Additionally, the subsequent research shows that the performance of face recognition is not only determined by the sparsity constraint (l1 regularizer), but also driven by the collaborative constraint (l2 regularizer). To overcome the issue mentioned above, in this paper, we propose an elastic net regularized dictionary learning based classification method. The proposed method is capable of improving the performance for face recognition according to class specific dictionary learning and elastic net regularizer. Moreover, to enhance the ability for handling nonlinear problems, we also extend the proposed method to arbitrary kernel space. Extensive experimental results on several face recognition benchmark datasets demonstrate the superior performance of our proposed algorithm.
CITATION STYLE
Wang, L., Wang, Y. J., & Liu, B. D. (2018). Elastic net regularized dictionary learning for face recognition. In Communications in Computer and Information Science (Vol. 819, pp. 140–150). Springer Verlag. https://doi.org/10.1007/978-981-10-8530-7_14
Mendeley helps you to discover research relevant for your work.