Palmprint Identification Based On Weighting Key Point Scheme

Biometrics is one of the most important and reliable methods for computer-aided personal identification. It regards its unique, reliable and stable physiological characteristics as the evidence of individual identity and applies the powerful computed ability of computer and technique of network to image processing and pattern recognition, then automated verify the individual identity.Biometrics combines the information technology with biology technology, which uses human inherent biological characteristics such as palm-print, iris, and face, and behavioral characteristics such as gait, signature and speech to confirm personal identity for replacing or strengthening the traditional personal identification approaches. And biometric technology has been applied to feasible systems in the market. The reliability of personal authentication is the key to the stringent security requirements in many application domains ranging from airport surveillance to electronic banking. At the same time, palmprint identification approach , regarded as one of the most up-and-coming biology technology, used its immutable and different traits to recognize the identity of persons. Also the palmprint identification approach has many physiological characteristics such as, uniqueness, stability, reliability.This thesis proposed a novel approach for personal identification based on weighting key point (WKP) scheme. In contrast with the existing approaches, this method can extract the local texture feature of palmprint more effectively. And based on WKP approach this thesis designed a palmprint identification system which can extract multimodal features, including hand shape and palmprint to facilitate the task of coarse-to-fine dynamic identification. That is to say, five hand geometrical features are used to guide the selection of a small set of similar candidate samples at the coarse level matching stage. In the fine level matching stage, multi-channel 2-D Gabor filters will be used to filter the ROI of candidate palmprint, and then, the weighting relative distance of key point approach is proposed to extract the local texture feature of palmprint. At last, the Mahalanobis distances, which are considered the most robust techniques, are used to measure the similarity between the feature of candidate samples and the template.In experiment, the channels of multi-channel 2-D Gabor filters, which can differentiate the sample more effectively, were chosen based on pattern match experiment. And then, according the weighting strategy proposed in this thesis, the key points in the corresponding channels were weighted. At last, the plentiful experimental results demonstrated the effectiveness and accuracy of these proposed method.