Συντάχθηκε 24-11-2011 09:11
από Eleni Stamataki
Email συντάκτη: estamataki<στο>tuc.gr
Ενημερώθηκε:
-
Ιδιότητα: σύνταξη/αποχώρηση υπάλληλος.
Τμήμα Ηλεκτρονικών Μηχανικών & Μηχανικών Υπολογιστών
ΠΑΡΟΥΣΙΑΣΗ ΔΙΠΛΩΜΑΤΙΚΗΣ ΕΡΓΑΣΙΑΣ
Ηλιού Δημήτριος
με θέμα
“Spectral Prediction From Filtered
Color CCD Cameras”
Παρασκευή 25 Νοεμβρίου 2011, 2:15μμ
Αίθουσα 2042, Πολυτεχνειούπολη
Εξεταστική Επιτροπή
Αν. Καθ. Κωσταντίνος Μπάλας (επιβλέπων)
Καθ. Μίνως Γαροφαλάκης
Επ. Καθ. Ευτύχιος Κουτρούλης
Abstract
Spectral Imaging is a powerful analytical tool and it has been used widely in a long list of different applications, from satellite imaging to biomedical diagnosis, non-destructive analysis etc. The current state of the art mainly includes, scanning spectral imaging system, which require either spatial or spectral scanning of the scene in order to collect the so called Spectral Cube. The Spectral Cube is a stack of tens of hundreds narrow band images. From the Spectral Cube, millions of spectra can be calculated and presented. Due to the time consuming data acquisition and processing procedures associated with the scanning hyperspectral Imaging Systems, their applications are restricted to static and stationary imaging conditions.
In this study we attempt to reduce that scanning and computational time of hyperspectral imaging by combining multispectral data acquisition with spectral estimation methods and algorithms. A number of algorithms were evaluated towards this end, and their performance in terms of prediction efficiency and calculation time was assessed. It was found that the the Wiener Estimation approach performed quite satisfactory. Particularly, it was found that with less than ten Spectral Bands, Wiener Estimation method predicted the spectra accurately.
These findings are setting the basis for the development of Real-Time HyperSpectral Imagers.
