الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 16 |  |  |
| | | from | 16 | | |
Abstract
Security has taken a great deal of people’s concern in the past few years. The world is developing fast and is driven towards full automation, the fact that urged the need for reliable security systems. Examples for applications where security systems are needed are: banking, A TM machines, passport control at airports, and other secured areas, where the person has to present his own identity for authorization. The search for a sign or code that provides a distinct and characteristic feature for each individual is a challenge for DSP researchers. Among other features, the iris of the human eye was found to be characteristic for each individual, even between genetically identical twins. Therefore, such biometric feature is used to design a fast and reliable security system.
In this thesis, a novel algorithm for iris isolation from the whole eye is proposed. Different tools have been used to extract representative iris features. These include the traditional DCT, the wavelet transform, and the Gabor wavelet transform. The generated iris codes are used to feed artificial neural network classifiers to authorize a set of persons having access to secured locations. Experiments have shown that the proposed feature extraction techniques have led to a recognition rate of above 90% on the average. However, the DCT technique has provided the best rate which is 96.2%. The thesis is organized as follows:
Chapter]. presents a brief review on biometrics and explains the anatomy of the eye. It is then concluded by presenting the aim of the thesis.
Chapter 2, describes the classical techniques dealing with iris pattern identification.
In chapter 3, the commonly used transforms, namely the discrete wavelet transform and the 2-D Gabor filters, are discussed in some detail.
A novel algorithm for isolating the iris by locating its inner and outer boundaries, is suggested and tested in chapter 4.
Chapler 5. presents the used feature extraction techniques namely; the 2-D Gabor tilters phase coefficients and their histogram, the discrete wavelet transform, and the discrete cosine transform .
In chapter 6, the classification results, using artificial neural networks, are presented and compared.
Finally, Chapter 7 concludes the presented work together with suggestions for future work .