الفهرس 
Abstract
Thesis Objectives and ContributionsOnly 14 pages are availabe for public view
 |  | from | 32 |  |  |
| | | from | 32 | | |
Abstract
AbstractIn the past decade, biomedical instrumentations have witnessed major developments and nowit has become very easy to measurehuman biomedical electrical signals. Moreover, new algorithms and techniques for biomedical signal processing have been developed. This opens the way for the use of the bio-electrical signals in different applications rather than the biomedical diagnosis. One of the most recent non-medical applications for these signals is the biometric authentication. This thesis studies the applicability of two bio-electrical signals, the Phono-Cardio-Gram (PCG) which is the electrical recordings of heart acoustics and Electro-Encephalo-Gram (EEG) which is the electrical recordings of brain waves, as biometric traits for human recognition.Such recognitionsystems traditionally provide two modes of functionality, identification and authentication (verification).Frameworks for human recognition using PCG and EEG signals are herein proposed and analyzed in bothscenarios.
For PCG signals, a framework is proposed for human recognition based on wavelet analysis. An automatic de-noising scheme for PCG signals is adopted based on discrete wavelet transform. Moreover, a filter bank structure is designed for feature extraction based on wavelet packet decomposition. The proposed system achieves higher recognition rates (in identification mode) and lower error rates (in verification mode) than the previously implemented approaches.
For EEG signals, an acquisition protocol is proposed based on eye blinking waveforms extracted from brainwaves. In this thesis, it is shown that eye blinking waveforms carry discriminate features that are capable for human recognition. Furthermore, a multi-level EEG based biometric system is developed using feature level fusion, where, features from eye blinking waveforms are fused with features extracted from brain waves during relaxation or visual stimulation protocols using canonical correlation analysis. The proposed multi-level EEG system using features from eye blinking outperforms the single-level EEG system, which is based only on relaxation or visual stimulation EEG, in the two modes of recognition; identificationand verification.