The Signature Verification Based On Mutli-biostatistics Features And Information Fusion

The traditional verification technology turn identification into validating the things that can identify a person, but this method has its inherent shortcomings. For example, the personal goods would be lost, the passwords would be forgotten or stolen by others, the traditional verification technology can't distinguish between the real users and the personates who get the real user's mark etc.Compared with the traditional verification technology, the biometrics verification has universality, stability and uniqueness. Identify based on biometrics has many advantages such as difficult to be forgotten, good capability of anti-counterfeit, difficult to be simulate or stolen, hand-carry and usable any time and place. According to analysis of several biometrics technologies, the applications of hand-writing signature verification are extensive and popular.Now, hand-written signature verification has been an important method of identification, and used in many fields. It is aimed at the traditional verification technology that we put forward an hand-writing signature verification model base on biometrics and information fusion:Firstly, the typical features of the coordinate and the curvature as well as the time information recorded were analyzed in the hand-written signatures, in the hand-written signature process 5 biometric features were summarized: the amount of zero speed in direction X and direction Y, the total time of the hand-written signatures, the total distance of the pen traveled in the hand-written process, the frequency of lifting the pen. And the raw data of biometric features vector is collected; establish an BP_GA sub-classifier based on BP neuron network geneticgene arithmetic and a RBF sub-classifier based on RBF neuron network, the raw data was disposed by normalization, the outcomes as inputs were imported into these sub-classifiers, then we can get the outputs; put forward a information fusion method based on support vector machine(SVM), construct an SVM of a third-order polynomial to mix the outputs from the sub-classifiers, and finally achieve the aim.The result of test indicates that the recognition rate (FAR) is obviously high (low) than signal classifier. The hand-written signature verification model based on information fusion can satisfy the application in common works and business contracts.