| 摘要: |
| 现有的基于稀疏表示的人脸识别算法在识别前需要将彩色人脸图像转换成灰度人脸图像，这样虽然提高了运算速度，但忽视了不同色彩通道数据本身所包含的信息及它们之间的相关性。为了利用不同通道间相关性，基于标签一致的K奇异值分解（LC-KSVD）字典学习算法，提出了一种适用于彩色图像人脸识别的字典学习算法。该算法将RGB通道数据顺序排列成列向量，并在稀疏编码的环节中，对正交匹配追踪（OMP）算法的内积计算准则进行修正，以此提高字典原子的色彩表达能力。在彩色人脸数据库上进行实验，结果表明：所提出的字典学习算法能够有效地提高识别率。 |
| 关键词: 彩色图像 人脸识别 稀疏表示 字典学习 稀疏编码 |
| DOI： |
|
| 基金项目:国家自然科学基金资助项目（61401108，61261023）；广西自然科学基金资助项目（2013GXNSFBA019272） |
|
| A dictionary learning algorithm for color face recognition |
| SHI Jinglan,CHANG Kan,ZHANG Zhiyong,QIN Tuanfa |
| () |
| Abstract: |
| The existing sparse-representation-based face recognition algorithms usually transform the color face images into gray images. Although this procedure increases the recognition speed,it ignores the information of the different color channels and the correlation among them. In order to utilize the correlation among different channels,based on the label consistent K-Singular Value Decomposition(LC-KSVD)algorithm,a new dictionary learning method for color face recognition is proposed. To improve the representing ability of each atom for color images,this algorithm concatenates R,G and B values into a single vector,and then introduces a new inner product into orthogonal matching pursuit(OMP)during sparse coding procedure. Experiments on different color face images datasets demonstrate that the proposed algorithm can achieve a higher recognition rate. |
| Key words: color images face recognition sparse representation dictionary learning sparse coding |
