基于镜像脸的FLDA单训练样本人脸识别方法

FLDA Used for Face Recognition with Single Training Image Per Person Based on Mirror Faces

Fisher线性判别分析(FLDA)是一种经典的基于特征提取的人脸识别方法。然而,在每类单训练样本时,FLDA无法对类内变化进行测量,因而无法使用。论文针对这一问题提出了一种新颖的解决方法,即通过利用每类已有的单个训练样本脸部图像获取其镜像图像,扩充原始训练样本集,从而解决原来在每类单训练样本情况下类内散布矩阵为零矩阵的问题。通过利用原始脸部图像和其镜像脸部图像计算出类内和类间散布矩阵,然后利用FLDA算法思想提取辨别性面部特征进而实现正确的分类和识别。实验结果表明,所提出的方法简单且高效,能够实现比现有方案更高的识别精度。

Fisher linear discriminant analysis(FLDA)is a classical face recognition method based on feature extraction.However,it cannot be used when each object has only one training sample because the intra-class variations cannot be statistically measured in this case.In this paper,a novel solution is proposed to this problem by using individual available face image to obtain its mirror face and integrate the original face image and its mirror face image to calculate the intra-class scatter matrix.Then the FLDA algorithm is used to extract the discriminant facial features to achieve the correct classification and recognition.The experimental results show that the proposed method is simple and efficient,and it can achieve higher recognition accuracy than the existing schemes.