Palmprint Recognition Method Based On Vector Representation

Biometric identification technology has achieved rapid development in recentyears, and traditional identification technologies (such as the key and card of doors,identity card and so on) can not satisfy some new requirements. The mainshortcomings of the traditional identification technology are shown as follows.They are not portable and can be copied. While the biometric identificationtechnology can avoid the above shortcomings. Biometric identification technologydirectly uses the physiological information (such as the face, palmprint and so on)and the behavior information (such as the voice and gait and so on) of individuals.These information is inborn, and they are not easily lost and copied.The focus of the dissertation is the appearance-based method. The palmprintimage is a pixel matrix and can be transformed into a column vector. Based onthese vectors, the test samples can be described by the training samples. Theappearance-based method has been used in face recognition successfully. In thisdissertation, this method and some improved methods are proposed. The improvedmethods are the appearance-based method in the feature space, palmprintrecognition method based on score level fusion and pamlprint recognition methodbased on adaptive fusion.The appearance-based method in the feature space transforms the samplesfrom the original space to the feature space, and then use the appearance-basedmethod in the feature space. Novelties of the method are the using of theappearance-based method in the feature space and the select of the parameter ofthe mapping function.Palmprint recognition method based on score level fusion fuses the localinformation extracted by the competitive coding method and the globalinformation extracted by the appearance-based method to recognize the samples.Novelties of the method are fusing of local information and global information.Pamlprint recognition method based on adaptive fusion uses2D palmprintinformation and3D palmprint information to recognize the samples. Novelties ofthe method are fusing of2D and3D palmprint images and the fusing way of them.