本文采用对称主元分析神经网络,用去冗余和权值正交相结合的方法对人脸进行特征提取和识别,具有所用特征数据量少,对测试图像进行特征提取运算量较小等特点.从而较好实现了大量人脸样本的存储和人脸的快速识别.模拟结果表明,在特征数量较少(20个)的情况下,图像仍可以较好地恢复和识别. In this paper, the symmetric principal component analysis neural network, the combination of de-redundancy and weight orthogonal feature extraction and recognition of the face, with the characteristics of the data used less, the characteristics of the test image feature less computation and so on So that the storage of a large number of face samples and the fast recognition of faces are achieved well.The simulation results show that the images can still be recovered and identified better with fewer features (20).