![]() | Only 14 pages are availabe for public view |
Abstract thesis presents a work in two main directions of the biometrics field; automatic speaker recognition and eye iris print recognition. This work has resulted in two recognition systems that are satisfactory reliable. The voice¬based recognition system yielded low FAR (False Acceptance Rate) and fRR (Valse RClcctio/1 Rate). The iris-pattern-based system yielded an accet)taule FRR, but a velY low FAR. A third system is then proposed combinillg these two biomeir;cs :11 n more ”effective” syste.,!1 yielding extraordinary results in ooth the ;~R,~ a,ld i .. e FAR. The thesis is organized in six chapters as follows: (’haptcr I, gIVes a brief introduction to the two biometrics use(; in t;.:s thesis as well as the pattern matching method (artificial neural networks) ~:lat was used for recognition. (~hapler 2, gives a brief introduClion 10 speaker recognition, ;Is principles, and classification as well as the methods used to accomplis:. recognition. At the end of the chapter possible applications, where S)eat Chapler 4, is organized in three patts. Part one sets the basics;or neurr .. networks from the detailed overview point of view of theory, architecture, and ;earning H.:es. A is presented for ~-perCCPlroni neurrtl network models. Part two gIves a brief overview for the fuzzy set theory. And prtrt three introduces the neuro-fuzzy systems. (’haptel’ 5, presents the detailed description of the two suggested recognition systems; one based on voice and the other on iris pattel1ls. A summary of the obtained results is then given followed by a comparison of both biometrics as reliable recognition systems. After that a description of the proposed hybrid model is given together with the obtained results. Chapter 6, discusses the suggested systems and the obtained ;esu:ts. And finrtlly some suggestions are given for future work. |