基于广义判别分析方法的人脸识别

提出基于广义判别分析的人脸识别方法，通过非线性核函数把样本映射到高维线性空间，然后在高维空间运用线性判决算法，从而获得输入空间非线性判决特征，可以很好地适应人脸图像中的光照、表情以及姿态等复杂的变化。实验证明该方法用较少的特征向量能获得比特征脸算法、Fisherfaces算法更高的分类准确率。

Face Recognition Based on Generalized Discriminant Analysis

The method based on Generalized Discriminant Analysis(GDA) is proposed for face recognition. Data points are mapped by means of nonlinear kernel function to high dimensional feature space to solve the problem in linear discriminant analysis algorithm, thus nonlinear characteristics of judgment can be available in input space, which is well adapt to facial illumination, expression, posture and other complicated changes. Experimental results show that GDA can get higher classification accuracy rate than Eigenfaces algorithm and Fisherfaces algorithm with less eigenvectors.