University of Piraeus (UNIPI)
Αλγόριθμοι ψηφιακής επεξεργασίας ηλεκτροεγκεφαλογραφήματος με στόχο την ταυτοποίηση του ατόμου
Abstract
dc:descriptionIn the present study the problem of developing an algorithm for the digital processing of an EEG was investigated for person identification purposes. According to this the the development of the following methodology is needed for; Research into the analysis of EEG signal in order to isolate those specific EEG segments that present stationary features. Experimental testing of many alternative digital - processing methods in order to isolate specific features from the aforementioned selected EEG segments. Classification of EEG segments according to the above features, using a different philosophy of classification method. statistical evaluation of the classification results in order to obtain sound conclusions concerning the reliability of the methods employed. The selection of the most efficient combination of feature extraction and classification / identification results , where its lowest value was generally greater than 75% while the highest value in specific cases was verging on 95%. In coclusion, the application of the proposed method lays the foundations upon which, in the future, successful person identification via an individual's EEG features, using digital EEG processing, may be reliably achieved
Degree
thesis:*- Grantor dc:publisher
- University of Piraeus (UNIPI)
- Year dc:date
- 2002
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Poulos, Marios
- Πούλος, Μάριος
Subjects
dc:subject × 16- Ατομική ταυτοποίηση
- Δίκτυα, Νευρωνικά
- Επεξεργασία σήματος
- Υπολογιστική γεωμετρία
- Αλγόριθμοι
- Ηλεκτροεγκεφαλογράφημα
- Person identification
- Signal processing
- Neuronic networks
- Computational geometry
- Algorithms
- Electroencephalogram (EEG)
- Φυσικές Επιστήμες
- Επιστήμη Ηλεκτρονικών Υπολογιστών και Πληροφορική
- Natural Sciences
- Computer and Information Sciences
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/13005