Sběr a analýza hyperspektrálních dat s využitím komprimovaného snímání
Loading...
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Tato bakalářská práce se zabývá komprimovaným snímáním a~jeho uplatněním při sběru hyperspektrálních dat. Cílem práce bylo seznámit se s metodou CASSI (Coded Aperture Snapshot Spectral Imaging), která umožňuje záznam hyperspektrálních dat pomocí jednoho snímku, provést pokusná měření i s následnými rekonstrukcemi obrazu a poté pro danou sestavu navrhnout doporučenou konfiguraci rekonstrukčních parametrů. Měření byla prováděna na soustavě HDES-VIS ve výzkumném centru TOPTEC, která pracuje v rozmezí vlnových délek od 400 nm do 900 nm (přibližně viditelné spektrum). Pro porovnávání různých přístupů rekonstrukce byly z naměřených dat vybrány tři typově odlišné scény, kde ke každé scéně byla následně vytvořena simulovaná data. Nejprve byla prováděna rekonstrukce pouze na simulovaných datech, díky čemuž byla odzkoušena funkčnost systému a mohl se snadno porovnávat vliv jednotlivých parametrů na proces rekonstrukce. Poté proběhla rekonstrukce i za použití reálných dat a ze získaných poznatků byla navržena ideální konfigurace jak pro~jednotlivé typy scén, tak pro celou soustavu obecně.
This bachelor thesis focuses on the compressed sensing and its usage in acquisition of hyperspectral data. The aim of this thesis was to get acquainted with the CASSI method (Coded Aperture Snapshot Spectral Imaging), which allows to record hyperspectral data using a single snapshot image, perform measurements with the image reconstruction and eventually suggest ideal configuration of reconstruction parameters. Measurements were performed with the system called HDES-VIS in the research centre TOPTEC, which operates in the spectrum range from 400 nm to 900 nm (approximately visible light). To compare different reconstruction approaches, three different types of scenes were selected from the measured data and for every type of scene simulated data were also created. At the beginning the reconstruction was performed only with simulated data, thanks to which the functionality of whole system could be tested and the effect of individual parameters on reconstruction could be easily compared. Subsequently, reconstruction with the real data was performed and from acquired knowledge the ideal configuration for every type of scene and also for whole system was suggested.
This bachelor thesis focuses on the compressed sensing and its usage in acquisition of hyperspectral data. The aim of this thesis was to get acquainted with the CASSI method (Coded Aperture Snapshot Spectral Imaging), which allows to record hyperspectral data using a single snapshot image, perform measurements with the image reconstruction and eventually suggest ideal configuration of reconstruction parameters. Measurements were performed with the system called HDES-VIS in the research centre TOPTEC, which operates in the spectrum range from 400 nm to 900 nm (approximately visible light). To compare different reconstruction approaches, three different types of scenes were selected from the measured data and for every type of scene simulated data were also created. At the beginning the reconstruction was performed only with simulated data, thanks to which the functionality of whole system could be tested and the effect of individual parameters on reconstruction could be easily compared. Subsequently, reconstruction with the real data was performed and from acquired knowledge the ideal configuration for every type of scene and also for whole system was suggested.
Description
Subject(s)
hyperspektrální snímání, komprimované snímání, metoda CASSI, HDES-VIS, hyperspectral sensing, compressed sensing, CASSI method, HDES-VIS