Using Phase and Directional Line Features for Efficient Palmprint Authentication

In palmprint recognition, researchers can extract ridges, singular points and minutia points as features from high resolution images while in low resolution images they generally extract principal lines, wrinkles and texture. And now almost all research concentrates on low resolution images for civil and commercial applications. This is also the focus of this paper. In this paper, a new method based on 2D hybrid Log-Gabor filter, D-S evidence theory and ANOVA (Analysis Of Variance) method is proposed for feature extraction. In palmprint images, 2D hybrid Log-Gabor filter is constructed by a number of weighted 2D Log-Gabor filters with different orientations to extract texture mixed-phase feature. And the integration of ANOVA method and D-S evidence theory is then employed for detecting directional line feature by dealing with the uncertain weak edges and the difficulty in threshold selection from multi-images. Finally, a decision based on the measurement of a modified Hamming distance is obtained. Experiments on the Polyu-Online-Palmprint Database show competitive performance.