一种对角LDA算法及其在人脸识别上的应用

2维特征抽取方法(如2DPCA、2DLDA),因为其抽取特征的速度和识别率要比1维的方法好,所以在人脸识别中得到了广泛的应用。最近基于2DPCA又提出了对角主成份分析(diagonal principal component analysis,DiaPCA),该方法由于保持了图像的行变化和图像的列变化之间的相关性,从而克服了2DPCA仅能反映图像行之间的变化,而忽略了图像列之间变化的缺点。但是,由于DiaPCA并没在特征抽取中融入鉴别信息,同时2DLDA也具有与2DPCA同样的缺点,从而分别影响了DiaPCA与2DLDA两种方法的识别性能。针对这一问题,提出了一种对角线性鉴别分析(diagonal linear dicriminant analysis,DiaLDA)的新算法,该新算法是基于对角人脸图像来求解最优鉴别向量。该新算法在ORL和FERET人脸库进行了实验,并与PCA、Fisherface、DiaPCA、2DLDA等方法进行了比较。实验结果表明,该方法比其他方法的识别性能要好。

A Diagonal Linear Discriminant Analysis Algorithm with Application to Face Recognition

Two-dimensional (2D)feature extraction using methods such as 2DPCA(two-dimensional principal component analysis)and 2DLDA(two-dimensional linear discriminant analysis)is of interest in face recognition because it extracts discriminative features faster than one-dimensional (1D)discrimination analysis.Recently,diagonal principal component analysis (DiaPCA)is proposed for face recognition based on 2DPCA.DiaPCA reserves the correlations between variations of rows and those of columns of images.It overcomes that the projective vectors of 2DPCA only reflect variations between rows of images and variations between columns of images are omitted,while the omitted variations between columns of images are usually also useful for recognition.However,DiaPCA in particular cannot make full use of discriminative information during process of feature extraction and the projective vectors of 2DLDA also only reflect variations between rows of images,Therefore recognition performance of DiaPCA and 2DLDA is affected.To solve the problem,diagonal linear dicriminant analysis (DiaLDA)was proposed in this paper.Experimental results on ORL and FERET face database demonstrate the proposed algorithm is superior to 2DLDA and DiaPCA method and some existing well-known methods.