Cai Jun, Chen Jing, Liang Xing
School of Optoelectronics, Beijing Institute of Technology, Beijing 100081, China.
Sensors (Basel). 2015 Jan 8;15(1):1071-87. doi: 10.3390/s150101071.
In this paper, a novel random facial variation modeling system for sparse representation face recognition is presented. Although recently Sparse Representation-Based Classification (SRC) has represented a breakthrough in the field of face recognition due to its good performance and robustness, there is the critical problem that SRC needs sufficiently large training samples to achieve good performance. To address these issues, we challenge the single-sample face recognition problem with intra-class differences of variation in a facial image model based on random projection and sparse representation. In this paper, we present a developed facial variation modeling systems composed only of various facial variations. We further propose a novel facial random noise dictionary learning method that is invariant to different faces. The experiment results on the AR, Yale B, Extended Yale B, MIT and FEI databases validate that our method leads to substantial improvements, particularly in single-sample face recognition problems.
本文提出了一种用于稀疏表示人脸识别的新型随机面部变化建模系统。尽管最近基于稀疏表示的分类（SRC）由于其良好的性能和鲁棒性在人脸识别领域取得了突破，但存在一个关键问题，即SRC需要足够大的训练样本才能获得良好的性能。为了解决这些问题，我们基于随机投影和稀疏表示，在面部图像模型中利用类内变化差异来挑战单样本人脸识别问题。在本文中，我们提出了一种仅由各种面部变化组成的改进型面部变化建模系统。我们还提出了一种对不同人脸不变的新型面部随机噪声字典学习方法。在AR、耶鲁B、扩展耶鲁B、麻省理工学院和FEI数据库上的实验结果验证了我们的方法带来了显著的改进，特别是在单样本人脸识别问题上。
