Research On Identity Verification Relied On Air Handwriting Signature

With the rapid development of information society,intelligent devices such as mobile phones,should implement the functions of communicating,online shopping and paying.The theft of identity can lead to serious society problems such as financial fraud,criminal activities and credit card fraud.To overcome these problems,some safety precautions should be made,one of which is to carry out the identity verification.At present,biometric authentication involving physiological characteristics is gradually being accepted by the society.However,biometric authentication relating to the behavioral characteristics is developing relatively slow.Compared to traditional on table two-dimension(2D)handwriting authentication,three-dimension(3D)air handwriting signature identity verification is a new technology for personal identity verification.It can provide a more friendly and natural way of human-computer interaction,and will become the research focus of identity verification in the future.In this paper,the 3D air handwriting Chinese signature is collected by accelerometer.The acceleration signals of Chinese signature written by the same person are similar and stable,the signatures of different people are different.Then,the identity can be verified through combining individual uniqueness with the matching algorithm based on dynamic warping.The main work of this paper can be described as follows.Signature preprocessing.In the initial stage of signing in the air,the hand may quiver due to the inadaptablity of human motion.Meanwhile,the human motion could cause the short stay in the last stage of signing in the air.The middle part of signature signal is the real information of human signing.In order to maintain the real signature information,the human hand would stay for short time in the initial and last stages.The collected handwriting signature is processed through three types of threshold interception algorithms.The best one of the threshold interception algorithms is acquired according to the interception performance.Through three order low pass Butterworth filter,the power frequency noise and artificial jitter noise can be filtered from the low frequency signature acceleration signal.The amplitude and the length of the signature signal all have differences.The size and length of the signature acceleration signal are structured to prepare for the following research of the effect on identity verification.Analysis of air signature identity verification global sequence matching algorithm.As the acceleration signal of Chinese signature is the time sequence,there exist many detailed differences.To be specific,the signature transition may be fast or slow,translation exists on the timeline and the signature lengths may be not the same.In this paper,taking account into these characteristics,we compare the Longest Common Subsequence algorithm,Global Sequence Alignment algorithm and improve the accuracy of the absolute value of the improved Dynamic Time Warping algorithm which all based on global sequence matching algorithm.Implementation of the Extreme GSA and improved DTW two-stage signature verification algorithm.When using global matching algorithm,the similarity of the signature acceleration data is accumulated through the maximum similarity score or minimum distance performance of global sampling point.Therefore,the time complexity of algorithm is high.In this paper,we propose the Extreme GSA and improved DTW two-stage structure algorithm for identity verification,it improves the efficiency of verification.The first stage extracts key extreme point of signal sequence,then perform GSA algorithm,it is fast;the second stage for the sample which has passed through the first-level certification.Using improved absolute value DTW algorithm for accurate verification.The advantage of taking the two-stage verification is that the combination of local feature points and global sampling points.The verification efficiency is improved,and the time complexity is reduced.Because there are subtle differences in the collection process of the air Chinese Signature,the further the effects of four different structured ways on verification are analyzed in the two-stage structure verification algorithm.The data of 920(23×40)Chinese signatures is collected from 23 people.The experiment results show that the best False Rejection Rate of 4.092%,False Acceptance Rate of 2.826%,Equal Error Rate of 4.092%.