Fingerprint recognition using principal Gabor basis function

Based on the Gabor transform, we propose a unified viewpoint for fingerprint image representation and recognition. In a local fingerprint image, we demonstrate that inter-ridge distances and ridge directions can be obtained by the principal Gabor basis function (GBF) directly. In addition, according to the locations of principal GBF on the spatial-frequency plane, the similarities of two local ridge structures can be measured easily. As compared with the global Gabor-based approach, the proposed method not only saves memory space, but also reaches a higher recognition rate