Detecting anomalies in remotely sensed hyperspectral signatures via wavelet transforms

An automated subpixel target detection system has been designed and tested for use with remotely sensed hyperspectral images. A database of hyperspectral signatures was created to test the system using a variety of Gaussian shaped targets. The signal-to-noise ratio of the targets varied from -95dB to -50dB. The system utilizes a wavelet-based method (discrete wavelet transform) to extract an energy feature vector from each input pixel signature. The dimensionality of the feature vector is reduced to a one-dimensional feature scalar through the process of linear discriminant analysis. Signature classification is determined by nearest mean criterion that is used to assign each input signature to one of two classes, no target present or target present. Classification accuracy ranged from nearly 60% with target SNR at -95dB without any a priori knowledge of the target, to 100% with target SNR at -50dB and a priori knowledge about the location of the target within the spectral bands of the signature.

Name thesis:degree_name

Master of Science (MS)

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Electrical Engineering

Grantor dc:publisher

University of Nevada, Las Vegas

Year

2000

Author dc:creator

• Morgan, Clifton Dennis

Contributors dc:contributor

• Lori Mann Bruce

Statement dc:rights

• IN COPYRIGHT. For more information about this rights statement, please visit http://rightsstatements.org/vocab/InC/1.0/

Language dc:language

English

Identifier

https://oasis.library.unlv.edu/rtds/1218

OAI identifier oai:identifier

oai:oasis.library.unlv.edu:rtds-2217