Monday, 08 April 2019 08:46


Hossam Emad Abdul Razzaq received his master's degree from the Department of Computer Science at the University of Technology for his thesis entitled:

 (An interdisciplinary mandate based on multiple biometric techniques)

Robust Authorization Based on Multi Biometric Techniques

In this thesis he proposed a multi-finger biometric identification system, which integrates the face and the shoulder veins at the level of characteristics. Both of the two fingers used are non-palms.

 The proposed system consists of two phases: the first is the processing phase, which contains a proposed algorithm for the discovery of the important palm region, and the second stage is a proposed symbiotic neural network structure consisting of two identical sub-networks with a common set of weights. The task is to compress the data and extract the properties, The two sub-networks are then merged to merge the properties. A new matching function has also been proposed.

Two fingerprint biometric systems were applied to one side and the other to the palm veins to compare and indicate the superiority of the proposed multi-finger biometric system.

The experimental results demonstrate the superiority of the proposed multi-fingerprint system, overcoming the disadvantages of both non-contact fingerprints and integration at the property level. It has also proven to be powerful against attack and more convenient for users. The system achieves a 99.10% high personal recognition rate and an error rate of 0.9009%. The system is very fast in matching, since the time required for each character identification process is 0.16 seconds. Thus, the proposed multi-fingerprint biometric identification system can be a powerful method of authorization.

The discussion committee consisted of: Ahmed Tarek and Dr. Abdul Amir Abdullah from the University of Technology / Computer Science Department and Prof. Bilal Ismail Khalil from Anbar University / Department of Computer Science and Student Supervisor Call of Fleih Hassan.



Read 1394 times Last modified on Monday, 08 April 2019 08:49