Summary: | 碩士 === 國立中央大學 === 資訊工程研究所 === 89 === The information technologies are now more and more blended into our daily life as the coming of electronic era. Traditional security ways, such as PIN code, are no longer reliable and difficult to satisfy the complex behaviors of e-commerce and information security. Unlike PIN code or passwords, biometrics can not be easily duplicated or stolen. Hence, they are more trustworthy and secure for identity verification.
There are various kinds of biometrics that can be utilized for personal identification. With the popularity of credit cards, signature plays a very important role in authenticity and authorization. People are getting more familiar with the process of handwritten signature verification. Most of the researches on signature verification are based on the real-stroke, or the trajectories that users write on a digitizer surface. However, forgery signatures written by skilled imitators usually challenge the performance of real-stroke-based verification systems. In this thesis, both feature extraction and verification methods are focus on the virtual-strokes. The virtual-strokes are the trajectories between each real-stroke in a signature. They are hard to be seen by naked eyes, but can be recorded by electronic devices. Due to this special characteristic of virtual-strokes, most of the forgery signatures can be easily identified. After the signature data has been acquired, each set of the one-dimensional signature data is converted into a two-dimensional signature image. Then, off-line signature verification methods are employed to accomplish the signature verification task. The developed system can prevent the occurring of forgery signatures to certain extend.
|