Feature extraction method for palmprint considering elimination of creases

A new method of extracting palmprint features is presented in this report. Since palmprint images have many creases which are organized like ridges, ordinary fingerprint feature extraction algorithms are unable to extract ridges. Consequently, the goal of this research is to construct a new feature extraction method which can extract ridges under these conditions. At first, the original palmprint image is divided into local images. The ridge candidates are extracted from each local image of the palmprint. Finally, only one candidate is selected as the ridge in the local image by estimating the continuity of certain properties. Experiments with palmprint images indicate that the proposed method can extract ridges better than conventional methods, especially in areas in which both ridges and creases exist.