In this work, hyperspectral images are images that capture n different wavelengths in the infrared spectrum, where 200 ≥ n ≥ 10, as opposed to the three visible light wavelengths captured in a standard image. We work with long wave infrared (LWIR) hyperspectral images, which inhabit the 8-15 µm wavelengths. The primary advantage of working with hyperspectral images is the ability to capture data that exists outside the visible spectrum, allowing for the extraction of data like invisible gases and anomalous harmful particles. With threats in the modern age that range from chemical attacks to radiation leakage, a method of isolating and extracting important features in hyperspectral images becomes a necessity.
The current attempts at extracting gas plume locations is to utilize a probability model and compares a dictionary of malicious gas plume signatures to all of the signals in a hyperspectral image. The main problem with this method is that it requires a dictionary of possible signatures, which may not be conceivable given that a synthetic gas can be created or an unknown gas can be used. I will utilize clustering methods, such as k-means and spectral clustering, instead to isolate the gas plume signatures and extract features of the landscape.
K-means successfully performs its job of obtaining useful information from a data set, being able to quickly demonstrate the better distance metric, as well as the viability of clustering methods to segment hyperspectral data. Due to the nature of k-means, it is not a candidate for consistently segmenting gas plume, as the results can vary greatly due to a small change in parameters. On the other hand, spectral clustering is able to accurately and consistently segment the gas plume. The open nature of spectral clustering also allows for the ability to further refine its results, allowing for segmentation of the gas to occur even when the gas becomes diffuse.
|Commitee:||Gao, Tangan, Lee, Chung-Min|
|School:||California State University, Long Beach|
|Department:||Mathematics and Statistics|
|School Location:||United States -- California|
|Source:||MAI 53/06M(E), Masters Abstracts International|
|Keywords:||Hyperspectral data, K-means clustering, Nystrom extension, Spectral clustering|
Copyright in each Dissertation and Thesis is retained by the author. All Rights Reserved
The supplemental file or files you are about to download were provided to ProQuest by the author as part of a
dissertation or thesis. The supplemental files are provided "AS IS" without warranty. ProQuest is not responsible for the
content, format or impact on the supplemental file(s) on our system. in some cases, the file type may be unknown or
may be a .exe file. We recommend caution as you open such files.
Copyright of the original materials contained in the supplemental file is retained by the author and your access to the
supplemental files is subject to the ProQuest Terms and Conditions of use.
Depending on the size of the file(s) you are downloading, the system may take some time to download them. Please be