Research On Several Key Aspects In Online Signature Verification

With the advent of the information age,the information revolution made a huge impact on the traditional style of office and life.In view of the more and more critical role of identity verification in the online payment and settlement,the electronic identity verification is becoming an important part of the information age.Meanwhile,with the great progress of biometric verification technology,biometrics as an identity authentication method has gradually attains to the requirements of automated,real-time and accurate,thus has attracted the widespread attention in the domain of pattern recognition.Commonly used biometrics includes fingerprint,face,finger vein,voice,signature etc.And signature,as an ancient and generally accepted biometrics,has been deeply studied and widely applied in the field of identity verification.In recent years,with the popularity and application of intelligent devices,the development of online signature verification technology which uses intelligent devices to acquire dynamic signature is making signature become an important method for identity verification again.The field of online signature has been developing for more than 50 years,but there are still a lot of room for improvement in the three key aspects: theory,application and evaluation standard.In this paper,we analyze,improve and expand the online signature verification system from three corresponding angles: algorithm improvement,application expansion and database standard construction.In theory research,this paper analyzes the most commonly used DTW algorithm in the field of online signature verification system,and introduces the main improvement of the DTW algorithm in an all-round way.Combined with these improvements,a more efficient multi distance measurement based MDB-DTW algorithm and forward and backward union FBUDTW algorithm are proposed.These improved algorithms get better results on the public database MCYT,SUSIG,and SVC2004,showing strong effectiveness and generalization.These algorithms are the most adaptable algorithms in the field of online signature verification,and they can provide a good algorithm foundation platform for future research.In application research,aiming at the problem of high dependence on device and low computational efficiency of existing algorithms in the field of traditional signature verification mode,combined with development of biometrics tends to the middle and long distance and non-contact verification mode,this paper first proposed the in-air signature verification mode in the world.The system realizes a set of software,hardware design and algorithm implementation,proving the validity and feasibility of the application of in-air signature from data acquisition method,pertinent algorithm improvement and quantitative,qualitative analysis.This research enriches the field of online signature verification from the perspective of identity verification mode and provides a new way of thinking for the application of online signature in real life.In the construction of evaluation standards,aiming at the loopholes in the lack of multimodal signature database and test protocol,this paper gives the first public multimodal signature database with corresponding test protocol and benchmark algorithm.The database consists of three subcorpora,namely,mobile,tablet and in-air signature subcorpus with a total of 2000 signatures.In total,50 subjects are included in each subcorpus,with 20 genuine samples and 20 skilled forgeries per subject.The database fills the gap in this field and provides a good experimental platform and sufficient experimental data for subsequent researchers to study and apply multimodal signature.Finally,we have summarized and prospected the research on online signature verification,and pointed out some problems that need to be solved in this field.Meanwhile,in order to facilitate the communication for researchers,part of the code and the whole database will be released in Github for researchers to learn or use.