Multispectral Palmprint Recognition Using a Hybrid Feature

Personal identification problem has been a major field of research in recent years. Biometrics-based technologies that exploit fingerprints, iris, face, voice and palmprints, have been in the center of attention to solve this problem. Palmprints can be used instead of fingerprints that have been of the earliest of these biometrics technologies. A palm is covered with the same skin as the fingertips but has a larger surface, giving us more information that the fingertips. The major features of the palm are palm-lines, including principal lines, wrinkles and ridges. Using these lines is one of the most popular approaches towards solving palmprint recognition problem. Another robust feature is the wavelet energy of palms. In this paper, we used a hybrid of these two features. Moreover, multispectral analysis is applied to improve the performance of the system. Main steps of our approach are: extracting principal lines and computing wavelet transform of the palm, computing block-based power of the resulting images as features, and finally making decision using a combination of the two features and a nearest neighbor criterion. The proposed method shows a best accuracy rate of 99.53% and an average accuracy of 98.8% on a database of 6000 palm samples.