Loose L 1/2 regularised sparse representation for face recognition
Loose L 1/2 regularised sparse representation for face recognition
复制标题
DOI：
10.1049/iet-cvi.2014.0114
复制
发表时间：
2015-04
期刊：
影响因子：
--
通讯作者：
Dexing Zhong;Zichao Xie;Yan-Rui Li;Jiuqiang Han
中科院分区：
文献类型：
--
作者：
Dexing Zhong;Zichao Xie;Yan-Rui Li;Jiuqiang Han
稀疏表示（或稀疏编码）已被应用于处理正面人脸识别。两种代表性方法是基于稀疏表示的分类（SRC）和基于协作表示的分类（CRC），其中查询人脸图像由所有训练样本的稀疏线性组合表示。 SRC和CRC的区别在于，前者利用编码的L 1-范数约束来保证稀疏性，而后者利用L 2-范数约束。在本文中，我们提出了一种用于人脸识别的松散 L 1/2 正则化稀疏表示（SR），称为 L 1/2 分类（LHC），其灵感来自于 L 1/2 正则化。此外，与原始算法相比，提出了迭代吉洪诺夫正则化（ITR）来有效地求解 LHC。使用 ITR，可以通过迭代来调整协作表示 (CR) 和 SR 之间的平衡。由于稀疏的L 1/2 正则化和迭代求解机制，LHC 可以获得更好的性能。对三个基准人脸数据库的大量实验表明，LHC 在处理正面人脸识别方面比最先进的基于 SR 的方法更有效。
Sparse representation (or sparse coding) has been applied to deal with frontal face recognition. Two representative methods are the sparse representation-based classification (SRC) and the collaborative representation-based classification (CRC), in which the query face image is represented by a sparse linear combination of all the training samples. The difference between SRC and CRC is that the L 1-norm constraint of coding is employed in the former to guarantee the sparse property, while the L 2-norm constraint is utilised in the latter. In this paper, we propose a novel loose L 1/2 regularised sparse representation (SR) for face recognition, named L 1/2 classification (LHC), which is inspired by L 1/2 regularisation. Additionally, an iterative Tikhonov regularisation (ITR) is proposed to solve LHC efficiently compared with the original algorithm. Using ITR, the balance between the collaborative representation (CR) and the SR can be tuned by the iterations. Attributed to the sparser L 1/2 regularisation and the iterative solution mechanism, a better performance can be achieved by LHC. Extensive experiments on three benchmark face databases demonstrated that LHC is more effective than the state-of-the-art SR-based methods in dealing with frontal face recognition.