Study On The On-Line Signature Verification Technique Based On Wavelet Analysis And Dynamic Time Warping

With the rapid development of the computer and communication technology, the great importance of information security is being emphasized. Personal identification is the essential prerequisite of the safety of the system. The traditional methods have their unchangeable disadvantages. However, the personal identification which is based on the biological features is becoming more and more widely used, for its specialty in dealing with all those disadvantages. On-line handwritten signature verification is one of the most important research areas in the field of personal identification based on biometrics technology. The technology is also regarded as a front subject in many fields such as pattern recognition and signal processing. An important advantage of the signature verification compared with other biometric attributes is its long tradition in many common commercial fields. On-line signature verification schemes extract signature features which characterize spatial and temporal characteristics of a signature. This thesis studies and proposes an on-line handwritten signature verification method based on wavelet analysis and dynamic time warping. The main contents of this thesis are as follows.The development of biometrics techniques is introduced at first, and a comparison of several kinds of biometrics techniques is taken. Research background of on-line handwritten signature verification and its development are briefly reviewed. A survey of existing signature verification methods is given, mainly discussing the latest progress. Key factors of research difficulties are concluded and some suggestions are given to future research. This thesis studies and proposes a blue print of the on-line handwritten signature verification's archetypal system, and analyzes data acquisition, preprocessing, feature extraction and feature matching in details. Then the equipment of data acquisition and the format of signature data are confirmed. All kinds of methods of preprocessing are analyzed, and the selection and extraction of signature features are studied. The function feature matching method is compared to the parameter feature matching method, and the superiorities of the former are narrated.The thesis focuses on the study of preprocessing and matching methods of thesignature signal. After eliminates the effective zeros and saves the location information of these effective zeros of the signal wave, proposes and implements twice segmentations towards to the signature signal. In order to divide the signature' strokes, the first segmentation are based on effective zeros. The second segmentation is based on the catastrophe points within the signature's stroke. We make use of Daubechies wavelet to carry out the second segmentation for the further processing. Finally, we utilize Dynamic Time Warping algorithm which is a dynamic programming technique to accomplish the signature signal's elastic matching. In the end, the experimental results are given. They show our proposed method is effective in the actual processing, so that it has good value for applications.