National Technical University of Athens (NTUA)
Ευφυείς πιθανοτικοί αλγόριθμοι αυτόνομης επισκόπησης εικόνων και εικονοσειρών
Abstract
dc:descriptionThe present doctoral thesis focuses on the problem of change detection in images and aims at supporting surveillance applications which employ intelligent image processing methods. In the presented analysis, the factors which require the employment of artificial intelligence algorithm for this specific process are discussed. Existing change detection algorithms rely on decision rules which require the availability of training image sequences. This restriction precludes their utilization when comparisons are restricted to two frames or when a time- varying noise model is observed. The contribution of the present thesis concerns the proposal of a novel statistic segmentation algorithm which performs the detection structural variations in frame pairs, without requiring the execution of a training stage on image sequences. After the description of the change detection problem in its general form, the presentation of two inspection algorithm variants follows. The first variant focuses on template-guided inspection of arbitrarily oriented targets, through the use of raster image templates. In order to implement the specific inspection application, the combination of the presented change detector and an image registration algorithm is proposed. A case study of undervehicle inspection is also presented. The second algorithm variant focuses on feature extraction from image sequences which exhibit time varying exposure conditions. Emphasis is placed on the development of highway surveillance algorithms for the support of Intelligent Transportation Systems. The exposure variability, which is observed in the related image sequences, requires the adaption of the utilized background model without the incorporation of static occlusions in its representation. In order to cope with this issue, the present thesis proposes the combination of the statistic segmentation algorithm with a novel complementary process which updates the background model and suppresses static occlusions.
Degree
thesis:*- Grantor dc:publisher
- National Technical University of Athens (NTUA)
- Year dc:date
- 2009
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Alexandropoulos, Theodoros
- Αλεξανδρόπουλος, Θεόδωρος
Subjects
dc:subject × 16- Ανίχνευση μεταβολών
- Εκτίμηση μοντέλου θορύβου
- Επισκόπηση βάσει προτύπων
- Χωρική ταύτιση εικόνων
- Επισκόπηση εικονοσειρών
- Ενημέρωση παρασκηνίου
- Change detection
- Noise model estimation
- Template-guided inspection
- Image registration
- Video surveillance
- Background update
- Επιστήμες Μηχανικού και Τεχνολογία
- Επιστήμη Ηλεκτρολόγου Μηχανικού, Ηλεκτρονικού Μηχανικού, Μηχανικού Η/Υ
- Engineering and Technology
- Electrical Engineering, Electronic Engineering, Information Engineering
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/17225