摘要：In this paper, a novel pre-processing algorithm is introduced to identify the principal lines froma palm-print image and a frequency domain feature extraction algorithm is then employed forpalm-print recognition, which can efficiently capture the spatial variations in the principal linesof a palm-print image. The entire image is segmented into several small spatial modules. Thetask of feature extraction is carried out in local zones using two dimensional discrete Fouriertransform (2D-DFT). The proposed dominant spectral feature selection algorithm offers anadvantage of very low feature dimension and it is capable of capturing precisely the detailvariations within the palm-print image. It is shown that because of the pre-processing step, thediscriminating capabilities of the proposed features are enhanced, which results in a very highwithin-class compactness and between-class separability of the extracted features. A principalcomponent analysis is performed to further reduce the feature dimension. From our extensiveexperimentations on different palm-print databases, it is found that the performance of theproposed method in terms of recognition accuracy and computational complexity is superior tothat of some of the recent methods.
