Online Signature Verification Based On Force Information

With the development of computer and network technology, biometric has become an increasingly important method for verifying an individual's identity. Handwritten signature verification is one of the main research areas of biometric. Since handwritten signature has long been used for identity verification in our daily life, it can be more naturally accepted by people comparing with other types of biometric attributes. Online signature verification can effectively reflect the individuality of signer because the dynamic information of handwriting is captured. Therefore it is the main research field of handwritten signature verification. Many researchers have devoted to the research of online signature verification, and numerous methods and approaches have been proposed. However, the recognition accuracy and reliability of current online signature verification system can't satisfy the requirements of some applications and should be enhanced further. One important reason is the inadequacy of writing force information captured using existing digital tablet. Supported by the NSFC project "force based acquisition and explanation of handwriting information", the acquisition of handwriting information based on force information and the method of online signature verification are deeply and systematically studied in this dissertation. The main works of this dissertation are as follows:Based on the detailed analysis of the work principle and deficiency of popular handwriting information acquisition instruments, a multi-axis force sensor based digital tablet was designed and developed. The tablet can collect signature trajectory and three dimension writing force simultaneously. The evaluation of the accuracy in determining the pen-tip position of the tablet was investigated, and a correction algorithm was proposed to estimate and reduce the systematic errors utilizing BP neural network.Considering the shape and writing force information of signature captured by the developed tablet, this dissertation studied the verification methods using local features and global features respectively.Firstly, considering the motor model of handwriting and the characteristic of Chinese signature, a signature is segmented using minimum writing velocity point.