Face recognition using discriminant sparsity neighborhood preserving embedding

详细信息	查看全文 | 推荐本文 |

• 作者：Gui-Fu Lu^a ; ^b ; ^{luguifu_jsj@163.com} ; Zhong Jin^a ; Jian Zou^a ; ^b
• 关键词：Sparse representation ; Dimensionality reduction ; Graph embedding ; Feature extraction ; Face recognition
• 刊名：Knowledge-Based Systems
• 出版年：2012
• 期刊代码：114_09507051
• 类别：et
• 出版时间：July, 2012
• 卷：31
• 期：Complete
• 页码：119-127
• 文件大小：1088 K

摘要

In this paper, we propose an effective supervised dimensionality reduction technique, namely discriminant sparsity neighborhood preserving embedding (DSNPE), for face recognition. DSNPE constructs graph and corresponding edge weights simultaneously through sparse representation (SR). DSNPE explicitly takes into account the within-neighboring information and between-neighboring information. Further, by taking the advantage of the maximum margin criterion (MMC), the discriminating power of DSNPE is further boosted. Experiments on the ORL, Yale, AR and FERET face databases show the effectiveness of the proposed DSNPE.