The Research Of Off-line Handwritten Signature Verification

The biometric identification is a kind of personal identification technique that recognizes status through the individual special physiology and the behavior features. Signature is one of the biological features used broadly for personal identification. And the handwritten signature personal identification technology has a wide application perspective in finance, law, office automation and engineering files. It is one of the most active and difficult problems in the biometric Identification.In this thesis, we firstly introduce the application background, the development history, and the trend of handwritten signature verification techniques at home and abroad. Secondly, we analyze the nature of the signature verification and the difficulties of the off-line signature verification. At last, we study thoroughly some key techniques and main algorithms of off-line handwritten signature verification.The main difficulty of off-line handwritten signature verification is feature extraction. In this thesis, we propose a method combining static and pseudo dynamic features which is based on wavelet transform. About the extraction of static features, we use the method of extracting the end points and the crossing points based on thinning of signature images. Regarding the pseudo dynamic feature, we extract high gray stable area feature based on histogram. And this is mainly concerned in this thesis. In the design of verification algorithms, we use two matching methods based on calculating distance for pseudo dynamic feature and two point features. We propose a matching algorithm of similarity for the feature of wave signal after wavelet transform. And these methods have the very good effect as a result of analyzing. Finally, we propose some ideas about the off-line Chinese signature verification.