Font Size: a A A

Online Signatue Verification And Its Applycation On Mobile Device

Posted on:2018-02-19Degree:MasterType:Thesis
Country:ChinaCandidate:L TangFull Text:PDF
GTID:2348330533466843Subject:Pattern Recognition and Intelligent Systems
Abstract/Summary:PDF Full Text Request
With the rapid development of smartphones hardware and the reduced cost of their manufacturing,smartphone penetration raising quickly.In addition,with the increasingly sophisticated function of software,many daily work and even important working assignments can be carried out on smart phones,which make people are more likely to rely on them.Thus,a secure and reliable user authentication on smartphones has already become an important issue for user.Authentication modes on smartphones include password authentication,fingerprint identification,iris recognition,online signature verification,speech recognition,etc.Among them,as a promising and effective solution,online signature verification on mobile device has caught more and more researchers' and companies' attention.According to its real application,online signature verification can be classified into local and cloud authentication.For the former,the whole system is running locally on a designated device as well as its all components.For the latter,enroll on a device while test on difference device,and the templates might be stored in cloud.According to the writing instruments employed in the signing process,online signature verification can be classified into two categories,finger-based and stylus-based signature.Compared with the latter,the former signature with index finger instead of a special stylus is more popular and more convenient in real application.So,this paper will study online signature verification algorithm and its application on mobile device for local authentication,and deign an online finger-based signature which in line with practical application.The main contributions of this paper as follows:(1)An improvement on matching model is given.To improve template matcher of online signature verification system,a novel matcher based on information divergence is proposed,which shifts the distance measurement between the test and reference samples to a similarity measurement between two distributions(generated between reference samples as well as between the test sample and all reference samples)and make full use of spatial information of reference samples.(2)An improvement on online signature verification algorithm have been investigated.By our in-depth study on DTW-based system,based on the novel matcher,a new DTW algorithm is proposed to exert the advantages of this new matching strategy.Finally,we given an in-depth analysis of several distance normalization methods and supplied some suggestion.(3)A study of online signature verification applications on mobile device was carried out.Firstly,we built a finger-writing online signature database and proposed an experiments protocol for it.Then,according to our research achievement on online signature verification algorithm,we designed an online finger-writing signature verification system which in line with practical application,and given some suggestion for its application,pointed out that the multi-sample can be used to generate the parameters of the model to increase the system performance,and the training problem of the matching model under the multiple training samples needs to be solved urgently.Finally,we designed and developed a file encryption software based on online signature authentication.Finally,experimental results on MCYT-100 and SUSIG signature databases show that the proposed online signature verification system achieved equal error rates(EERs)of 2.25% and 1.70% when 10 reference samples were used and 3.16% and 2.13% when 5 reference samples were used respectively,which illustrates the effectiveness of the proposed strategy compared with other state-of-the-art strategies.The results also show that the online finger-writhing system can achieve 6.7% and 7.1% of EER on our self-build online finger-writing signature database,with 478 ms and 314 ms of average time-consumption when 8 and 4 reference samples can be used,which illustrates that the proposed system is in line with application.
Keywords/Search Tags:Online signature verification, Dynamic Time Warping, Information divergence, Local authentication, File encryption, Mobile device
PDF Full Text Request
Related items