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
https://orcid.org/0000-0002-8926-1789
AccessType
Open Access Dissertation
Document Type
dissertation
Degree Name
Doctor of Philosophy (PhD)
Degree Program
Electrical and Computer Engineering
Year Degree Awarded
2024
Month Degree Awarded
February
First Advisor
Mario Parente
Second Advisor
Marco Duarte
Third Advisor
Janice Bishop
Subject Categories
Data Science | Numerical Analysis and Scientific Computing | Remote Sensing | Theory and Algorithms | The Sun and the Solar System
Abstract
The Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) has proven to be an invaluable tool for the mineralogical analysis of the Martian surface. It has been crucial in identifying and mapping the spatial extents of various minerals. Primarily, the identification and mapping of these mineral spectral-shapes have been performed manually. Given the size of the CRISM image dataset, manual analysis of the full dataset would be arduous/infeasible. This dissertation attempts to address this issue by describing an (machine learning based) automated processing pipeline for CRISM data that can be used to identify and map the unique mineral signatures present in a CRISM image. The pipeline leverages a highly discriminative representation learned through the use of Generative Adversarial Networks, such that in this novel representation space simple distance metrics are sufficient to discriminate between even very similar spectral shapes. The pipeline leverages this enhanced feature space to set up an open set classification problem that labels each new pixel as either a member of a known mineral class or novel spectral shape (or outliers). Following this, a segmentation technique is used on the outliers to group them, and further, reduce them to a representative set of the novel spectral shapes present in the image. These novel spectral shapes can then be labeled based on expert analysis and used to update the open-set classifier. The performance of these tools are validated over a subset of CRISM images from different parts of the Martian surface such as Jezero Crater, North East Syrtis, and Mawrth Vallis.
DOI
https://doi.org/10.7275/36490689
Recommended Citation
Saranathan, Arun M., "Automated identification and mapping of interesting mineral spectra in CRISM images" (2024). Doctoral Dissertations. 3089.
https://doi.org/10.7275/36490689
https://scholarworks.umass.edu/dissertations_2/3089
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Included in
Data Science Commons, Numerical Analysis and Scientific Computing Commons, Remote Sensing Commons, Theory and Algorithms Commons, The Sun and the Solar System Commons