Palmprint Recognition Methods Based On Half-orientation Features

In the rapid development of an information-based society,the need for effective identification of people's identities has become more and more urgent.In the existing palmprint recognition method,the palmprint recognition method based on the orientation characteristics is the most prominent and has achieved a very good recognition effect.This method generally uses the Gabor filter to filter palmprint images in each directions to obtain palmprint orientation feature representation.However,this method uses the Winner-take-all rule to extract the orientation of the filtered maximum response value as the orientation characteristic of the ridge line,and often ignores many other features that have stronger distinguishability on the palmprint.On the other hand,the method considers that the palm lines are straight lines,ignoring the presence of a large number of curv e features on the palm lines.Therefore,an effective method for extracting more distinguishable orientation features and curve features on palmprint is of great significance for improving palmprint recognition.According to the existing methods for extrac ting the orientation features of palmprint ridges,based on the method of half-orientation coding for extracting the characteristics of palmprint curves,in order to obtain more distinctive characteristics of the curve direction,this paper proposes the fo llowing two research work: 1)A new method for extracting palmprint half-orientation features is proposed.This method judges the extracted two half-orientation features,replacing one of the halforientation in the same orientation,so that the two extracted half-orientation features are completely differently,the half-orientation feature can completely cover the palm curve features,so as to obtain more rich distinguishing features.2)Propose a palmprint half-orientation feature extraction method combining fragility,by analyzing the instability of the palmprint on the different images of the same palmprint when the filter response value of the filter is low,by filtering the instable features allow the acquisition features to have stronger feature expre ssion capabilities.To verify the effectiveness of the palmprint recognition method based on the halforientation feature proposed above,this study selected the public palmprint data set of the Hong Kong Polytechnic University to conduct experiments on pa lmprint verification and palmprint identification,and on the basis of traditional orientation-based features.Compared with the method proposed in this paper,the accuracy of palmprint recognition has been improved after obtaining more palm curve feature s in the half-orientation feature extraction method.The equal error rate is reduced by 8.6%,and the half-orientation feature extraction method is combined with fragility.After further enhancing the expressiveness of the features,the error rate dropped by a maximum of 11.6%.The recognition error rate was lower than that of the halforientation coding method,and a better palmprint recognition effect was achieved.Experiments show that the recognition accuracy of palmprint recognition has been further improved after extracting more distinctive features and curve features.