Date of Award:
Doctor of Philosophy (PhD)
Electrical and Computer Engineering
Jacob H. Gunther
As a result of its capacity for material discrimination, hyperspectral imaging has been utilized for applications ranging from mining to agriculture to planetary exploration. One of the most common methods of exploiting hyperspectral images is spectral unmixing, which is used to discriminate and locate the various types of materials that are present in the scene. When this processing is done without the aid of a reference library of material spectra, the problem is called blind or unsupervised spectral unmixing. Independent component analysis (ICA) is a blind source separation approach that operates by finding outputs, called independent components, that are statistically independent. ICA has been applied to the unsupervised spectral unmixing problem, producing intriguing, if somewhat unsatisfying results. This dissatisfaction stems from the fact that independent components are subject to a scale ambiguity which must be resolved before they can be used effectively in the context of the spectral unmixing problem. In this dissertation, ICA is explored as a spectral unmixing approach. Various processing steps that are common in many ICA algorithms are examined to assess their impact on spectral unmixing results. Synthetically-generated but physically-realistic data are used to allow the assessment to be quantitative rather than qualitative only. Additionally, two algorithms, class-based abundance rescaling (CBAR) and extended class-based abundance rescaling (CBAR-X), are introduced to enable accurate rescaling of independent components. Experimental results demonstrate the improved rescaling accuracy provided by the CBAR and CBAR-X algorithms, as well as the general viability of ICA as a spectral unmixing approach.
Stites, Matthew R., "Assessing and Enabling Independent Component Analysis As A Hyperspectral Unmixing Approach" (2012). All Graduate Theses and Dissertations. 1199.
Copyright for this work is retained by the student. If you have any questions regarding the inclusion of this work in the Digital Commons, please email us at .