Font Size: a A A

Research On Feature Extraction Algorithm Of Online Handwritten Signature Authentication

Posted on:2018-06-01Degree:MasterType:Thesis
Country:ChinaCandidate:W WangFull Text:PDF
GTID:2428330545455808Subject:Computer system architecture
Abstract/Summary:PDF Full Text Request
With the development of science and technology,the Internet has brought huge improvements to people's lives.People enjoy this convenient service also triggered a new round of personal information security issues,how to protect the security of personal information has become the focus of attention.Due to the traditional way of identity authentication cannot meet the needs of modern society,researchers have begun to research based on biometric identity.Because of its flexibility and convenience,online handwritten signature attracts most researchers.In order to improve the performance of online handwritten signature authentication system,two different signature authentication schemes are proposed:multi-signature signature template based on discrete Frechet distance and DTW signature based on stable segmentation.The multi-feature combination template based on discrete Frechet distance mainly introduces the theory of curve similarity,converts the signature feature data into a signature characteristic curve,and judges the authenticity of the signature according to the similarity of signature characteristic curves.Firstly,signature features are extracted from the signature data and other types of features are calculated according to the basic features.Such as the X,Y direction of the linear velocity can be calculated by the signature line speed,displacement and so on.Secondly,according to the characteristic curves of each signature,the maximum and minimum points of each characteristic curve are found out.By the signature sample training,the discriminant threshold of each signature is calculated.Finally,the signature template is generated according to the orthogonal experiment design,and the signature of each signature is compared with the signature of each corresponding template,and the false positive rate and false positive rate of all signatures are calculated.Compare the error rate of the original template and the combined template,and finally select the optimal combination signature template to obtain the multi-signature combination template.DTW signature verification based on stable segmentation is to analyze the stability of the pen segment in the signature data,extract the stable signature pen segment,and use DTW to solve the authentication process.Mainly divided into the following steps:First,according to the maximum speed of the sample signature segmentation processing,get the signature of different segmentation.The second step is to compare the multiple real signatures of the users,select the signed handwriting segments of the common trend to perform the signature sample training,and calculate the signature discrimination threshold.Research shows that the handwriting segments of these common trends are stable,and the use of these stable signature stroke segments can effectively improve the operation efficiency and accuracy of the signature system.The third step is to find the DTW distance for each segment of signature,and to sum the segment DTW distances to calculate the DTW distance of the stable segments in the entire signature.Finally,the signature obtained through signature training threshold to determine the authenticity of the signature.The signature database used in this thesis is SVC2004,which analyzes and preprocesses the signature data set in the database.Experiments show that the signature authentication system based on discrete Frechet distance multi-signature template is more stable and has higher authentication efficiency than the signature authentication system randomly selected fixed signature template.The DTW signature authentication based on stable segmentation removes the unstable factors in the signature and reduces the computation of the signature data,which improves the system authentication speed and improves the system authentication efficiency.Therefore,the multi-signature signature template authentication based on discrete Frechet distance and the DTW signature verification based on stable segmentation can all improve the signature verification system and provide a new idea for the improvement of the signature system.
Keywords/Search Tags:Biometrics, Online handwritten signature certification, Discrete Frechet distance, Dynamic Time Warp
PDF Full Text Request
Related items