Off-campus UMass Amherst users: To download campus access dissertations, please use the following link to log into our proxy server with your UMass Amherst user name and password.
Non-UMass Amherst users: Please talk to your librarian about requesting this dissertation through interlibrary loan.
Dissertations that have an embargo placed on them will not be available to anyone until the embargo expires.
Author ORCID Identifier
Open Access Dissertation
Doctor of Philosophy (PhD)
Electrical and Computer Engineering
Year Degree Awarded
Month Degree Awarded
Marco F. Duarte
Frank P. Seelos
Data Science | Numerical Analysis and Computation | Numerical Analysis and Scientific Computing | Signal Processing | Theory and Algorithms | The Sun and the Solar System
Hyperspectral imaging has been deployed in earth and planetary remote sensing, and has contributed the development of new methods for monitoring the earth environment and new discoveries in planetary science. It has given scientists and engineers a new way to observe the surface of earth and planetary bodies by measuring the spectroscopic spectrum at a pixel scale.
Hyperspectal images require complex processing before practical use. One of the important goals of hyperspectral imaging is to obtain the images of reflectance spectrum. A raw image obtained by hyperspectral remote sensing usually undergoes conversion to a physical quantity representing the intensity of light energy, called radiance. In order to obtain the reflectance spectrum of surface, the contribution of atmosphere needs to be addressed and then divided by a spectrum of ``white reference.'' Furthermore, the obtained reflectance spectra of image pixels are likely to be the mixtures of multiple species due to limited spatial resolution from orbits around planets.
Hyperspectral unmixing is an attempt to unmix those pixels - to identify substantial components and estimate their fractional abundances. Hyperspectral unmixing has been widely explored in the literature, but there are still many aspects yet to be studied. The majority of research focuses on the development of methods to retrieve correct substantial components and accurate fractional abundances. Their theoretical aspects are rarely investigated. Chapter 2 will pursue a theoretical aspect of sparse unmixing, one of the hyperspectral unmixing problems and derive its theoretical conditions that guarantee the correct identification of substantial components.
Hyperspectral unmixing can also be used for other stages of hyperspectral data processing. Chapter 3 explores the application of hyperspectral unmixing to the processing of hyperspectral image acquired by the Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) onboard the Mars Reconnaissance Orbiter (MRO). In particular, new atmospheric correction and de-noising methods for the CRISM data that use a hyperspectral unmixing to model surface spectra, are introduced. The new methods remove most of the problematic systematic artifacts present in CRISM images and significantly improve signal quality.
Chapter 4 investigates how hyperspectral images acquired from orbits can be combined with ground exploration. In the recent rush of the launch of many Martian ground rover missions, it is important to effectively integrate knowledge obtained by hyperspectral remote sensing from orbits into ground exploration for facilitating Martian exploration. In specific, this dissertation solves the problem of matching hyperspectral image pixels obtained by the CRISM with ground mega-pixel images acquired by the Mast Camera (Mastcam) installed on the Curiosity rover on Mars. A new systematic methodology to map the CRISM and Mastcam images onto high resolution surface topography is developed.
Itoh, Yuki, "Hyperspectral unmixing: a theoretical aspect and applications to CRISM data processing" (2022). Doctoral Dissertations. 2636.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Data Science Commons, Numerical Analysis and Computation Commons, Numerical Analysis and Scientific Computing Commons, Signal Processing Commons, Theory and Algorithms Commons, The Sun and the Solar System Commons