As the human face image is at a high dimension, how to reduce the dimension quickly and extract the main features of the human face images was studied in this paper, as well as how to achieve efficiency in human face recognition. The first step was to study 2DPCA, LDA and RBF neural network algorithm. Then, the 2DPCA+LDA method was designed for human face image feature extraction and dimension reduction, as well as human face image classification, taking advantage of RBF neural network method. Finally, this method was used in the experiments based on ORL face database and it was proved that high recognition rate could be reached and the time of human face recognition could be reduced well.%本文针对人脸识别应用中存在的人脸图像的高维特点,对人脸图像如何快速降维以及提取人脸主要特征,同时实现人脸识别的高效性问题进行研究。首先,分别研究了2DPCA、LDA、RBF神经网络算法。然后,综合设计了应用2DPCA+LDA方法对人脸图像进行降维和特征提取,同时利用RBF神经网络方法对人脸图像进行分类识别。最后,将本方法在ORL人脸数据库上进行实验验证,证明在识别率和时间上得到了很好的效果。
展开▼