Design Of Online Handwriting Signature Verification Used In Smartphone

With the development of Internet, besides communication, smartphone alsocarries the tasks of network communication, online entertainment, online shoppingand mobile payment. Because smartphone contains personal information,communication records and transaction information, the loss of smartphone will leadto the leakage of personal information and economic losses. Therefore, smartphonesecurity management has become a key issue of mobile Internet technology.Handwriting signatures can reflect the signers’ unique writing behaviorcharacteristics. Signatures are convenient, easy to accept, difficult to forge and unique.As an effective verification method, signatures are widely used in legal documents,contract, authorization management and other verification occasions. Smartphonesand tablets can collect finger or pen signature trajectory. The handwriting signatureverification relied on the unique characteristics of signature, can be used forinformation security management of mobile Internet. This security management canprovide higher level security and extensive application.Collecting signatures on capacitive touch screen is different with traditionaltablets. The acquisition methods, sampling rate and resolution ratio will cause greatinfluence on signature verification. For the characteristics of low resolution ratio, lowsampling rate and non-uniform time sampling, this paper studies pretreatmentmethods. Contents include:Signature normalization: rotate signatures to horizontal direction. And thennormalize the size and position, so that all signatures are in the same range andlocation. Redundant points’ deletion: in the process of pen up and pen down, there aretwo or three random repeated points. So we should delete them. Signatureinterpolation resampling: Smartphone sampling rate is about45points per secondwhich is lower than tablet. And the signature sampling is non-uniform. We fit thetrack of original signature better and distribute equal intervals according to timesequence by resampling.The computing power of intelligent terminal is poorer than PC. Signatureverification methods require both effective and low computing power. Two methodsare used in this paper. One is characteristic parameter method. Signature features areextracted and then the signatures are certified by certain rules. The other is signaturetrajectory similarity judgment method. The modify rematch dynamic time warping is proposed. There are four steps. Firstly, we divide signatures into strokes by extremepoints. Secondly, the strokes are matched by dynamic time warping (DTW). Thirdly,these wrong matching strokes are modified. Fourthly, we use DTW in each stroke.The proposed algorithm is tested, and the verification rate is above92.975%for386signatures come from9persons.