Comparison of PCA, LDA and GDA for palmprint verification

In this paper, we have compared use of PCA (Principal components analysis) with two powerful feature extraction techniques LDA (Linear discriminant analysis) GDA (Generalized discriminant analysis) which have already been used in palmprint verification. For testing purpose 10 colorful whole-hand images of each hand of 43 persons are collected by a digital camera, namely, a small dataset of 860 images is built. The experimental results show that the best verification result is obtained with the GDA based method, whose average minimal total error rate is only 0.11% on the dataset.