Application Of Palmprint In Identity Identification And Authentication

As an important complement of the current recognition technologies by using biometric features, the research on palmprint identification is very meaningful. It is a challenging task to complete the feature extraction.On the basis of referring to arithmetic of other biometric technologies, three arithmetics are presented at the stage of extracting feature:(1)It analyzes palmprints by using a multi-resolution method-wavelet transform, and defines a palmprint texture feature(Wavelet Energy Feature, WEF). According to inherent different resolutions of primary-line and wrinkles, the thickest and widest primary-lines are analyzed in low resolutions, while the thinnest and slenderest wrinkles are analyzed in high resolutions.(2)It uses a set of Gabor filter to convolute the palmprint at different scales, extracts the average and deviation of every sub-band as the features, and absolutization distance is used to weigh the difference among features. On the basis of Gabor texture feature and referring to the hierarchical retrieval, the global energy feature is extracted at the first level by using Gabor filter, and local texture feature is extracted at the second level by using LBP operator in order to realize two levels of retrieval.(3)It presents a method of palmprint identification by fusing energy and direction. First, a set of Gabor filter is applied to the preprocessed palmprint at different scales and different directions, and then divided the sub-band into fuzzy blocks. Calculate the energy of every small block, find the small block with maximum energy at the same scale and different directions, and mark the Gabor filter's direction of this block with maximum energy. Using Gray code to decode direction feature, calculate the Eculidean distance of energy feature and the hamming distance of direction feature, and finally these two differentdistances are fused by weight.This paper discusses all performances of every arithmetic, and an off-line system of palmprint recognition is realized by Matlab on the basis of Gabor texture feature. By comparing with the performances of every arithmetic, it provides the research direction and experience for the further research.