Image-Recognition For Identification And Authentication In Electronic Transactions

The development of network technology offers Internet finance an unprecedented wave of development,accompanied by the emergence of a variety of security issues.An endless stream of viruses,Trojans,fishing sites,telecommunications fraud has brought a huge threat to the user’s property security.To this end,the network authentication has become an important means of network payment transactions to ensure transaction security.Traditional authentication methods are authentication,bank authentication verification,etc.,although it can be a good proof of identity,but in the need to require more stringent verification features,especially how to identify and authenticate the trader in network authentication is extremely important.Identification based on image,especially face image,has become an important innovation direction for recognition and authentication.First of all,the thesis designs the business methods for authentication and proposes a more rigorous method,face verification,according to the background demand.Secondly,the thesis puts forward the software structure of the identification and authentication system according to business needs,designs the corresponding module and implements the module function.Finally,the thesis tests the system and analyzes the results.This thesis uses the training model of open source face recognition engine(SeetaFace)to realize face detection,feature point location and feature extraction and comparison,and realize the face recognition function on this basis.Blink detection uses the template matching to the locate the eyes,according to whether the center of the eye’s coordinates in the current connected area to determine whether the tester is blinking.The result of system test shows that according to the different illumination,face angle and image resolution,the accuracy rate of the face classification in the standard image library is 98.6%,which shows a good discrimination degree.The main contents of the thesis are as follows: 1)On the basis of existing verification methods,the thesis putted forward the concrete way of using the face image to identify and verify,and realized the corresponding function;2)In the electronic transaction process,in order to prevent illegal users to use static photos posing as legitimate users,the thesis proposed and realized based on real-time video blink detection verification,to ensure that the identified object is real person;3)Usually in the financial charge scene,the process of certification often requires two legitimate users at the same time to complete the authorization.Therefore,the thesis proposed and implemented the authentication method based on double recognition,that is,to identify and verify double people’s video or photograph;4)Designed and implemented the identification and authentication system;5)Tested the identification and authentication system and analyzed the results.