摘要
提出了一种新的有监督核局部线性嵌入算法(SKLLE),并将算法应用于面部表情识别中。该算法通过非线性核映射将人脸图像样本投影到高维核空间,然后将人脸图像局部流形的结构信息和样本的类别信息有效地结合进行维数约简,提取低维鉴别流形特征用于表情分类。SKLLE算法不仅能发现嵌入了高维人脸图像空间的低维表情子流形,增强了局部类间的联系,而且对新样本有较好的泛化性。基于JAFFE面部表情库的实验结果表明,该方法能很好地实现维数约简,达到最高识别率(100%)所需的鉴别维数仅为二维,有效地提高了面部表情识别的性能。
A novel supervised kernel local linear embedding （SKI.I.E） method is introduced to facial expression recognition, which maps face images to a high dimensional kernel space through nonlinear kernel mapping, then fuses prior class-label information and nonlinear facial expression submanifold of real face images to extract discriminative features for expression classification. SKLLE can not only gain a perfect approximation of facial expression manifold, and enhance local within-class relations, but also can do well on the new samples. The experimental results on JAFFE database show that the proposed method can achieve the highest recognition rate （100%） using only 2D embedding feature vectors, which improves face expression classification performance effectively.
出处
《光学精密工程》
EI
CAS
CSCD
北大核心
2008年第8期1471-1477,共7页
Optics and Precision Engineering
基金
重庆市自然科学基金资助项目(No.CSTC2006BB215)
关键词
流形学习
核技巧
局部线性嵌入
有监督学习
面部表情识别
manifold learning
kernel trick
Local Linear Embedding （LEE）
supervised learning
facial expression recognition
