الفهرس 
CHAPTER 1: INTRODUCTIONOnly 14 pages are availabe for public view
 |  | from | 32 |  |  |
| | | from | 32 | | |
Abstract
Traditional approaches, such as fingerprinting and face recognition, are becoming more and more susceptible to counterfeiting. As a result, new and more counterfeit proof biometrics modalities have been considered, one of them being the heartbeat pattern acquired by an electrocardiogram (ECG). ECG signal is one of the new techniques for human identification. Because the ECG signal varies from person to person, it can be used as a new biometric for individual identification.
The main objectives of this thesis is to present a novel personal identification system that uses the ECG signals and ECG images. The proposed system is divided into four main phases: pre-processing phase, feature extraction phase, classification phase and identification phase. For ECG images a real database is created and used, it consists of 120 ECG images which is gathering from 20 persons. For ECG signals, famous databases like MIT/BIH arrhythmia database and Abdominal ECG database are used.
The preprocessing phase is applied on the ECG image. Preprocessing should remove all variations and details from an ECG image that are meaningless to the identification method. For feature extraction phase, detecting all waves that formed ECG signal is very important for human identification. This thesis proposes a new algorithm for detecting most of ECG signals such as the signals that have positive R-peaks, the signals that have negative R-peaks and the signals that have positive and negative R-peaks. In classification phase ECG is classified into normal and abnormal signals. The proposed system is worked in normal ECG signal. In identification phase, Support Vector Machine and Neural Network algorithms are used to achieve person’s identification.
The performance of the ECG identification system is evaluated by calculating the sensitivity (Se) and the positive predictivity (+P). The proposed system achieves high sensitivity results for preprocessing ECG images, extracting ECG features and for human identification. The proposed algorithm was used on 50 ECG images the result of feature extraction algorithm is 92.36% and 97.94% for sensitivity (Se) and positive predictivity (+P) respectively, 98.4% identification rate using Feed-forward Neural Network algorithm is obtained and 98.81% identification rate using SVM algorithm (RBF kernel) is obtained. The proposed algorithm was used on 25 ECG records from MIT/BIH arrhythmia database, the accuracy rate is 99.97%. The proposed algorithm was used on 25 ECG records from fetal database, the sensitivity and the positive predictivity are 100% and 96.78% respectively. At the end, identification system is built. The proposed system achieves high sensitivity results for extracting ECG features and for human identification.