全局和局部特征相融合的人脸识别算法

摘要

In order to further improve the correct rate and efficiency of face recognition, a novel image recognition algo-rithm based on global and local features exaction is proposed. The local scatter matrix and global scatter matrix are intro-duced to represent global features and local features, and optimization problem is established based on identical samples as close as possible while heterogeneous samples as far away as possible, support vector machine is used to construct face classifier, and the simulation experiments are carried out on three face databases. The results show that, compared with other face recognition algorithms, the proposed algorithm has improved recognition accuracy and improved recognition effi-ciency of face images.%为了进一步提高特征提取效率和人脸识别正确率，提出一种融合全局和局部特征的人脸识别算法。引入局部散度矩阵和全局散度矩阵，两者分别表征样本的全局特征和局部特征；基于同类样本尽可能的紧密而异类样本尽可能远离的事实，构造最优化问题，采用支持向量机建立人脸分类器，并通过仿真实验测试算法的性能。仿真结果表明，该算法不仅提高了人脸识别正确率，而且提高了人脸识别效率。