Palmprint verification using wavelet packet

In this paper, a palmprint, which mainly contains principal lines, wrinkles and ridges, is regarded as a texture. A palmprint texture has intrinsic multi-resolution characteristic in different directions. We analyze palmprint textures using wavelet packet and define a novel palmprint feature, called wavelet packet feature (WPF), for personal authentication. WPF can reflect the characteristic of a palmprint texture at different resolutions and in different directions, thus it can efficiently describe the palmprint. In the verification tests, 98.5% accuracy has been obtained by the proposed approach with 4-level WPF. The experimental results demonstrate the power of the proposed approach.