University of Twente Student Theses
Bayesian Approach on Hyperspectral Unmixing of Sandstone Rock Using Hapke's Model
Lukman, M.N. (2024) Bayesian Approach on Hyperspectral Unmixing of Sandstone Rock Using Hapke's Model.
|
PDF
4MB |
| Abstract: | Spectroscopic measurements have long been utilized to analyse the electromagnetic absorption patterns of minerals, aiding in the identification and quantification of mineral abundance. An important challenge in this field is the occurrence of mixed pixels, where multiple minerals are captured as one pixel. Spectral unmixing techniques aim to resolve this by decomposing the mixed pixel spectra into their individual mineral components. This research evaluates hyperspectral unmixing techniques for identifying and quantifying mineral compositions in sandstone core samples, focusing on the use of the Hapke Model to address the non-linearity of particulate surface spectra and Bayesian Spectral Unmixing (BSU) to model the uncertainty related to spectral unmixing using the Markov-Chain Monte Carlo (MCMC) method. We also examine the impact of additional spectral processing methods—first derivative, hull removal, and normalization—on the accuracy of spectral unmixing. Furthermore, the BSU method is compared to the Fully Constrained Least-Square (FCLS) method to evaluate its effectiveness. The study began with experiments using computer-generated mixtures and mineral powder mixtures, followed by spectral unmixing on the core sample. Our results indicate that the simplified Hapke Model significantly enhances the accuracy of spectral unmixing compared to the Linear Mixing Model (LMM). The BSU method provides more detailed information on the variability and uncertainty of mineral abundances, outperforming FCLS in scenarios with complex mineral mixtures. Additionally, spectral processing methods, such as normalization, were shown to reduce spectral variability and improve unmixing results. |
| Item Type: | Essay (Master) |
| Faculty: | ITC: Faculty of Geo-information Science and Earth Observation |
| Subject: | 38 earth sciences |
| Programme: | Geoinformation Science and Earth Observation MSc (75014) |
| Link to this item: | https://purl.utwente.nl/essays/101556 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page