Font Size: a A A

Research On Online Signature Verification Based On Curvature

Posted on:2018-07-05Degree:MasterType:Thesis
Country:ChinaCandidate:H TanFull Text:PDF
GTID:2428330596454648Subject:Computational Mathematics
Abstract/Summary:PDF Full Text Request
With the progress and development of science and technology,the problems brought by network and information security are becoming more and more serious,the signature as a behavioral characteristics,with non-invasive,and will not be lost,forgotten,easily accepted,is a recognized identity technology.It has a wide range of applications in the field of information security,which has very important research value and significance.In recent years,the signature verification technology has achieved great development.However,there are still some cases where the signature verification method is not ideal for some complex signatures.Therefore,the research of signature verification technology needs to be continued.In many signature authentication methods,the coordinate information of the curve is not very good for the translation and rotation transformation.Therefore,this paper focuses on the signature authentication of coordinate information,and proposes an improved coordinate information signature authentication method and two-level authentication mechanism,Compared with the original coordinate information signature verification methods,the proposed methods respectively improve the recognition rate of the curve under the effect of translation and rotation transformation.The main work and innovation of this article are as follows:Firstly,This article firstly studies the principle and process of handwritten signature authentication,and realizes the data acquisition,preprocessing and feature extraction in the process of signature verification through numerical examples,which lays the foundation for the subsequent algorithm design and matching decision.Secondly,Arming at the shortcomings of the robustness of the coordinate information signature authentication method for translation and rotation transformation,this article proposes an improved coordinate information signature authentication method.The method includes improvements in two aspects: Firstly,the curvature of the extreme point is calculated by using the original two-dimensional coordinate information.Curvature represents the degree of curvature of the curve,with the invariance of translation and rotation,the proportion of telescopic characteristics.Secondly,Combined with the pressure characteristics,the curvature and torsion of the extreme points are calculated by using the original three-dimensional coordinate information.The torsion characterizes the degree of distortion of the curve and has the invariability of the rigid body,and thus obtains two kinds of features with stronger description ability.Finally,the validity of the improved coordinate information signature verification method for the translation and rotation transformation is verified by analyzing the experimental results.Thirdly,Aiming at the problem that the signature is easy to be forged in the writing process,this paper proposes a new two-level signature authentication mechanism.The first level authentication is based on the signature curve matching of discrete Fréchet distance and curvature.Using the improved coordinate information signature authentication method to extract the one-dimensional feature,the discrete Fréchet distance is taken as a distance measure,and the relationship between the two points is defined,then the minimum Fréchet distance of the peaks point curvature and the minimum Fréchet distance of the trough point curvature between the two curves are calculated.The absolute value of the difference between the two is not affected by the translation and expansion,and then compared with the set threshold to identify the true signature.This level authentication is to quickly eliminate the random forged signature.The second level authentication is based on the signature curve matching of Hausdorff distance and curvature and torsion.Using the improved coordinate information signature authentication method to extract the two-dimensional feature,the Hausdorff distance does not need to establish the one-to-one correspondence between the points.By calculating the maximum distance between the two points,the similarity between the curves is described,then the eight-dimensional distance matrix corresponding to the two signature curves is calculated,and the final matching rate is obtained by training and verifying again through the neural network.This level authentication is to identify the skilled forged signature.
Keywords/Search Tags:signature verification, curvature, torsion, discrete Fréchet distance, Hausdorff distance
PDF Full Text Request
Related items