摘要： 提出一种基于局部Fisher鉴别分析(LFDA)和优化支持向量机(SVM)的高效人脸识别算法。在综合考虑局部几何结构和类别信息的基础上，利用LFDA将高维人脸数据映射到低维特征空间，避免维数灾难问题。在该低维特征空间中，使用经乘性更新规则训练的优化SVM对人脸数据进行分类识别。在人脸数据库上的实验结果表明，该算法的运算速度较快，识别准确率较高。

关键词: 人脸识别, 局部Fisher鉴别分析, 支持向量机, 流形学习, 特征提取, 乘性更新规则

Abstract: To effectively deal with face recognition problem, an efficient face recognition algorithm based on Local Fisher Discriminant Analysis(LFDA) and optimal Support Vector Machine(SVM) is proposed in this paper. By comprehensive consideration of local geometric structure and class information, the high dimensional face data are first mapped into lower dimensional feature space with LFDA so that the curse of dimensionality can be avoided, the optimal SVM that is trained by the multiplicative update rule is used to classify and recognize different face image herein. Experimental results on face image databases demonstrate that the proposed algorithm not only has fast running speed, but also achieves higher recognition accuracy.

Key words: human face recognition, Local Fisher Discriminant Analysis(LFDA), Support Vector Machine(SVM), manifold learning, feature extraction, multiplicative update rule

中图分类号: 

• TP181