人脸识别技术是基于生物特征的识别方式，与指纹识别等传统的识别方式相比，具有准确、隐蔽和非侵扰等特性，较容易被用户接受，因此人脸识别技术在诸多领域都有广泛的应用。人脸识别技术中的人脸特征提取是近年来基于生物特征研究的热点之一。 本文在总结了人脸识别技术的内容和方法的基础上，具体讨论了三种基于子空间分析的特征提取方法：主成分分析PCA方法、线性判别分析LDA方法和独立成分分析ICA方法，详细阐述了这三种特征提取方法的主要思想、算法流程及实现方法，并分别对这三种方法在ORL人脸数据库上进行了仿真分析，讨论了影响算法识别率的主要因素，实验结果表明ICA方法相对于另外两种方法具有较高的识别率。文中阐述了Gabor小波变换技术及二维Gabor小波变换在人脸特征提取中的应用，利用Gabor小波在表达人脸局部特征上的优势，提出了一种结合二维Gabor小波变换和ICA的人脸识别方法，并对算法进行改进，将人脸图像用Gabor小波的幅值和相位表示，将幅值与相位信息作为识别的依据，用ICA方法计算得到分离矩阵，用最近邻分类器分类识别。该方法很好的将两者的优点结合起来，实验结果表明，改进算法具有很高的识别率，尤其是在训练样本数量较少的情况下，识别率仍保持在90%以上，具有一定的实用价值。

Face recognition technology is a recognition method based on biological features. Compared with fingerprint recognition and other traditional means of identification, it is easier to be accepted by users because of its accuracy, concealment and non-intrusive features. Just for this reason, face recognition technology has a wide range of applications in many fields. In recent years, facial feature extraction in face recognition technology has become one of hot spot based on biological characteristics. Based on summary of the content and method of face recognition technology, this paper discusses in detail three types of feature extraction method based on subspace analysis: PCA (principal component analysis) method, LDA (Linear Discriminate Analysis) method and ICA (Independent Component Analysis) method. Then we elaborate the main idea of the three methods in detail and introduce their algorithm process and implementation. Meanwhile, we conduct experiment analysis of the three methods on the ORL face database and discuss the main influencing factors of recognition rate of the three algorithms. The experimental results show that ICA method has a higher recognition rate than the other two. In this paper, we present two dimensional Gabor wavelet transform and its application in extraction of facial features. Gabor wavelet has a big advantage on expression of local features of face. By making use of this advantage, a face recognition method is proposed combining two dimensional Gabor wavelet transform with ICA, which improves the algorithm. Face image is represented with the amplitude and phase of Gabor wavelet, and the amplitude and information of the phase are considered as the basis of recognition. Separating matrix is evaluated by ICA method, and recognition according to their types is realized by nearest neighbor classifier. This method combines both the advantages of Gabor wavelet transform and ICA method. Experimental results show that the improved algorithm has a high recognition rate, and it still maintains high recognition rate, more than 90%, especially in cases with small training samples, which is favorable for practical application.