Automated eXpert Spectral Image Analysis
Abstract
AXSIA performs automated factor analysis of hyperspectral images. In such images, a complete spectrum is collected an each point in a 1-, 2- or 3- dimensional spatial array. One of the remaining obstacles to adopting these techniques for routine use is the difficulty of reducing the vast quantities of raw spectral data to meaningful information. Multivariate factor analysis techniques have proven effective for extracting the essential information from high dimensional data sets into a limted number of factors that describe the spectral characteristics and spatial distributions of the pure components comprising the sample. AXSIA provides tools to estimate different types of factor models including Singular Value Decomposition (SVD), Principal Component Analysis (PCA), PCA with factor rotation, and Alternating Least Squares-based Multivariate Curve Resolution (MCR-ALS). As part of the analysis process, AXSIA can automatically estimate the number of pure components that comprise the data and can scale the data to account for Poisson noise. The data analysis methods are fundamentally based on eigenanalysis of the data crossproduct matrix coupled with orthogonal eigenvector rotation and constrained alternating least squares refinement. A novel method for automatically determining the number of significant components, which is based on the eigenvalues of the crossproduct matrix, hasmore »
- Authors:
- Publication Date:
- Research Org.:
- Sandia National Laboratories (SNL), Albuquerque, NM, and Livermore, CA (United States)
- Sponsoring Org.:
- USDOE
- OSTI Identifier:
- 1230701
- Report Number(s):
- AXSIA V2.0 BETA Ver; 001428IBMPC01
- DOE Contract Number:
- DE-AC04-94AL85000
- Resource Type:
- Software
- Software Revision:
- 01
- Software Package Number:
- 001428
- Software Package Contents:
- Media Directory; Software Abstract; Media includes Source Code, Executable Module(s), Compilation Instructions, Installation Instructions, User Guide, Linking Instructions\1 CD Rom
- Software CPU:
- IBMPC
- Open Source:
- No
- Source Code Available:
- Yes
- Related Software:
- Intel Math Kemel Library V6.1, Intel IPP library V3.0, Mathlab V6.5.1
- Country of Publication:
- United States
- Subject:
- chemical analysis, data processing
Citation Formats
Keenan, Michael R. Automated eXpert Spectral Image Analysis.
Computer software. Vers. 01. USDOE. 25 Nov. 2003.
Web.
Keenan, Michael R. (2003, November 25). Automated eXpert Spectral Image Analysis (Version 01) [Computer software].
Keenan, Michael R. Automated eXpert Spectral Image Analysis.
Computer software. Version 01. November 25, 2003.
@misc{osti_1230701,
title = {Automated eXpert Spectral Image Analysis, Version 01},
author = {Keenan, Michael R.},
abstractNote = {AXSIA performs automated factor analysis of hyperspectral images. In such images, a complete spectrum is collected an each point in a 1-, 2- or 3- dimensional spatial array. One of the remaining obstacles to adopting these techniques for routine use is the difficulty of reducing the vast quantities of raw spectral data to meaningful information. Multivariate factor analysis techniques have proven effective for extracting the essential information from high dimensional data sets into a limted number of factors that describe the spectral characteristics and spatial distributions of the pure components comprising the sample. AXSIA provides tools to estimate different types of factor models including Singular Value Decomposition (SVD), Principal Component Analysis (PCA), PCA with factor rotation, and Alternating Least Squares-based Multivariate Curve Resolution (MCR-ALS). As part of the analysis process, AXSIA can automatically estimate the number of pure components that comprise the data and can scale the data to account for Poisson noise. The data analysis methods are fundamentally based on eigenanalysis of the data crossproduct matrix coupled with orthogonal eigenvector rotation and constrained alternating least squares refinement. A novel method for automatically determining the number of significant components, which is based on the eigenvalues of the crossproduct matrix, has also been devised and implemented. The data can be compressed spectrally via PCA and spatially through wavelet transforms, and algorithms have been developed that perform factor analysis in the transform domain while retaining full spatial and spectral resolution in the final result. These latter innovations enable the analysis of larger-than core-memory spectrum-images. AXSIA was designed to perform automated chemical phase analysis of spectrum-images acquired by a variety of chemical imaging techniques. Successful applications include Energy Dispersive X-ray Spectroscopy, X-ray Fluorescence Spectroscopy, Laser-Induced Fluorescence Spectroscopy and Time-of-Flight Secondary Ion Mass Spectroscopy.},
doi = {},
url = {https://www.osti.gov/biblio/1230701},
year = {Tue Nov 25 00:00:00 EST 2003},
month = {Tue Nov 25 00:00:00 EST 2003},
note =
}
To receive further information, fill out the request form below. OSTI staff will begin to process an order for scientific and technical software once the signed site license agreement is received. You may also reach us by email at: .