Search

Browse

• Communities
• Authors
• Year
• Titles
• Subjects

My Space

• Login

Development of spectral data estimation methods towards expanding the data dimensionality in spectral imaging

Vassalos Konstantinos

Full record

---

Summary

HyperSpectral imaging is a technology that provides detailed spatial and spectral information about an object, combining spectroscopy with imaging capability. It captures detailed spectral information across the electromagnetic spectrum, obtaining the spectrum for each pixel in an image and enabling precise material identification and analysis. However, the high dimensionality of HyperSpectral data often poses significant challenges in terms of data acquisition, storage, and processing, particularly when aiming to achieve high spatial and spectral resolution simultaneously. This thesis focuses on the development of advanced HyperSpectral data estimation methods to enhance spectral resolution, reconstruct missing information, and expand the effective dimensionality of HyperSpectral datasets. To achieve this, we propose and evaluate computational techniques that leverage machine learning, signal processing, and mathematical modeling to improve HyperSpectral data estimation. These methods aim to reconstruct incomplete spectral signatures, enhance spatial and spectral resolution, and mitigate data acquisition limitations. The proposed approaches are tested on both synthetic and real HyperSpectral datasets, assessing their effectiveness in terms of accuracy, robustness, and computational efficiency.

Available Files

Services