Palmprint Feature Extraction Based on Curvelet Transform

An approach of palmprint feature extraction based on the second frequency band curvelet coefficients is proposed in this paper. As we all know, the veins of the palmprint are the important factors to recognize the palmprint. Because curvelet transform can effectively presents the lines and curvilinear structure, we perform the curvelet transform to the palmprint image. By analyzing the features, amplitude spectrum and energy statistics of the curvelet coefficients for each frequency band, we can see that the curvelet coefficients in the second frequency band can represent the veins of the palm better, so we choose them as the palmprint features. Firstly, the palmprint image is transformed by the fast discrete curvelet transform via wrapping to extract the second frequency band coefficients as the palmprint features. Then, the PCA method is used to reduce the dimension of the extracted features and get a more representative palmprint features. Finally, we use the nearest neighbor method to classify the palmprint. The results of the experiments performed in PolyU 2D palmprint database show that our approach has the high effectiveness and robustness in palmprint recognition.