Research On Assistant Localization Of Palmprint And Design Of Authentication System On IOS Platform

Traditional authentication methods on mobile terminals are mainly knowledge-based methods,such as numeric password,nine-grid graphic password,which are easy to be forgotten,cracked and peeked.Biometrics,including physiological and behavioral characteristics,are used for novel authentication to enhance authentication security,stability and reliability on mobile terminals.The existing identity authentication methods based on fingerprint,face,etc,have some shortcomings.Compared with these biometric modalities,palmprint has many advantages,including rich discriminant information,less restrictive conditions,less leaking,low cost and so on.However,the current palmprint verification on mobile terminals suffers from several severe technique challenges,like uncontrollable location and gesture of hands,complex background,variant illumination,limited hardware resources,to name a few.The main research contributions and achievements of this dissertation are as follows:(1)Summary of the knowledge systematization of palmprint verificationThe shortcomings of the existing biometrics on mobile terminals and the advantages of palmprint verification are introduced.In addition,the feasibility of the palmprint verification on mobile terminals is analyzed.We survey the recent research on palmprint verification technologies,and point out the shortcomings and unsolved problems.The performance evaluation indices of palmprint verification are elaborated in details.(2)“Double-point” assistance for acquisition and localizationIn order to overcome the uncontrollable problems of the location and posture of hands,“double-point” assistance is designed for acquisition and localization.During capture stage,the two bottom-points of the valleys,which are between the index finger and the middle finger,the ring finger and the little finger,are located on two“assistance points”,i.e.the centers of the two “restricting boxes”.“Double-point”assistance can effectively restrain the location and posture of acquired hand,reduce the complexity of the follow-up image preprocessing,improve the execution efficiency and accuracy.(3)“Twice adaptive skin-color model” for palm segmentationThere are a large number of interference regions,whose color,shape,and texture are similar to those of palm,in the background of the palmprint images taken with amobile terminal in complex scenarios.Traditional skin-color models cannot meet the accuracy requirements.To solve this problem,a “twice adaptive skin-color model”based on adaptive Gaussian skin-color model is developed to segment palm region.The pixels in palm region within restricting boxes are used to train “twice skin-color model”,so the correct rate and accuracy are remarkably improved.(4)Improved ASM(active shape model)for palm segmentationTraditional ASM methods combine gray information and shape information,but they are not good in complex scenes,including mobile environments.An improved ASM method is developed,in which Perux method normalizes the shape of palm.Then the shape model of the palm is calculated with principal component analysis.Finally,the color likelihood is used to replace the gray information in traditional methods for target fitting.The improved ASM method reduces the complexity,while improves the accuracy and robustness.(5)Design and implementation of palmprint verification system on iOS platformA palmprint verification system is developed on iOS platform.The palmprint images for enrollment and authentication are captured with the build-in camera.The algorithms in this dissertation are implemented and tested on this system,which confirms the effectiveness of the proposed algorithms.To sum up,we review the research on the state-of-the-art palmprint verification methods on mobile terminals,and propose “double-point” assistance for acquisition and localization.On the basis of this assistance scheme,two novel palm segmentation algorithms,namely “twice adaptive skin-color model” and “improved ASM”,are designed.The works in this dissertation reduce the complexity of preprocessing,and also improve the accuracy and robustness.Thus this dissertation has very important theoretical significance and application value.