الفهرس | Only 14 pages are availabe for public view |
Abstract Biometrics refers to the entire class of technologies and techniques to uniquely identify humans. Two types of biometrics methods can be considered; first, physiological biometrics, which involves data derived from the direct measurement of some part of the human. The other, behavioral biometrics, involves data derived from an action taken by a person, or indirectly measures characteristics of the human body. In each biometrics systems, feature detection is of utmost importance. Once the main features have been extracted, a decision network (classification network) can be build with relative ease. When the feature detector network performs poorly, however, it is extremely difficult, often impossible, to construct accurate classification. Thus, the extracted feature should have a reasonable degree of robustness (invariance) against shift, rotation, scale and other variation due to individuals. This thesis uses global feature extraction method (wavelet transform) to prevent these problems by introducing two techniques. One of them is off-line handwritten signature verification system. The other one is Human iris verification system. Each system uses Discrete Wavelet Transform (DWT) in extracting the robust feature, Wavelet Neural Network (WNN) as a classification technique and Genetic Algorithms (GA), to search for more robust and integrated parameters that characterize each (signature/human iris) pattern. |