Research And Implementation Of Video Handwriting Signature Verification Algorithm Based On The HMM/ANNs Hybrid Model

Nowadays, with the rapid development of Internet technology, the problems brought by cyberspace and information become deteriorated, which make complexity and harmfulness more obvious. How to distinguish the identity of a person accurately and effectively becoming very important. Online signature verification is one of the most effective biometric-based identity- authentication methods. The design and implementation of an online signature verification system involve data acquisition, feature extraction, feature selection, decision making, and performance evaluation. At present, signature verification technic is not that perfect and its liability and accuracy is lower than using inherent physiological characteristics of human body such as fingerprint, DNA, and iris.In this paper, the video handwriting authentication method and application of signature are studied, the main contributions are as the following:Firstly, this thesis presents the design and implementation of a camera-based, humancomputer system for acquisition of handwriting. The camera focuses on the sheet of paper and the process of the handwriting; computer analyses the resulting sequence of images to track the trajectory of the pen and determine the starting and ending times when the pen is in contact with the paper. The trajectory is shown to have sufficient spatio-temporal resolution and accuracy to enable handwritten character recognition.Secondly, an approach based on HMM/ANNs hybrid is presented for online signature verification. The hybrid HMM/ANNs model is constructed by using a type of time delayed Neural Networks as local probability estimators for HMM.We circumvent this issue by employing a group of ANNs,but not a single ANNs to predict the probability of each state on condition of the current observation and the previous state,and each ANNs is corresponding to a state.Then, the training of the model will be more effective.Finally, some experiments with this method are made on the signature database in the Matlab development platform. The result indicates that the proposed method can get better performance such as lower ERR compared to HMM.