Online Signature Verification Based On Feature Function Method

With the rapid development of network and communication technologies, there comes an urgent and higher requirement on the verification, which should be more veracious, safe and practicable. For those applications under the network environment, especially in the-commerce related field, it has become a key technique problem to decide a person's identity swiftly, conveniently and accurately.Handwritten signature verification is a new identity verification method. It can is more safe, reliable and convenient compared to the conventional ways. Signature verification can be classified into online verification and offline verification, according to whether the signature sample is captured during the writing process. And it can be sorted into function method and parameter one according to the features used in verification. Thanks to the support of National Natural Science Foundation of China (NSFC), we have developed the F-Tablet, a signature information capture device, for the project "Handwritten information capture and explanation based on Force". Furthermore, the online signature verification based on feature function method has been studied using the F-Tablet, which can capture the three dimension writing forces.The work of this dissertation can be summarized as follows:Firstly, we devised the F-Tablet. Compared to the existing signature capture devices, the F-Tablet is capable of obtaining not only signature shapes, but also writing forces and some other dynamic information during the writing process. A signature database was constructed based on the obtained information. And this database could provide signature data for the performance analysis of different verification algorithms.Secondly, we put forward an improved DTW algorithm based on the signature information captured by the F-Tablet. The experiments were carried out with Linear Distance Algorithm, DTW algorithm and improved DTW Algorithm using different signature information. The validity of the improved DTW algorithm is proved by the experimental results. An iterative experiment is designed to decide the weight of different informationThirdly, we put forward a multi-step verification algorithm with segmentation referring to stable waves and wave matching. In this algorithm, the first and second...