Real time gamma-ray signature identifier

• US 20070108379A1
• Filed: 03/27/2006
• Published: 05/17/2007
• Est. Priority Date: 03/25/2005
• Status: Active Grant

• Current Assignee: Lawrence Livermore National Security LLC (Government of the United States of America)
• Original Assignee: Regents of the University of California (University of California)
• Inventors: Wong, James, Rowland, Mark, Perkins, Dwight, Ham, Cheryl, Gosnell, Tom
• Granted Patent: US 8,180,579 B2
• Time in Patent Office: Days
• US Class Current: 250/260
• CPC Class Codes: G01T 1/36 Measuring spectral distribution of X-rays or of nuclear radiation spectrometry pulse selection circuits per se H03K; investigation of materials by radiation diffraction G01N23/20; spectrometer tubes H01J49/00

Abstract

A real time gamma-ray signature/source identification method and system using principal components analysis (PCA) for transforming and substantially reducing one or more comprehensive spectral libraries of nuclear materials types and configurations into a corresponding concise representation/signature(s) representing and indexing each individual predetermined spectrum in principal component (PC) space, wherein an unknown gamma-ray signature may be compared against the representative signature to find a match or at least characterize the unknown signature from among all the entries in the library with a single regression or simple projection into the PC space, so as to substantially reduce processing time and computing resources and enable real-time characterization and/or identification.

Claims

1. A real-time method of identifying radioactive materials by their gamma-ray signatures comprising:

providing at least one library of predetermined spectra for various nuclear material types;

performing principal component analysis on the library to produce a corresponding transformed basis set representing the library and indexing each spectrum of the library in principal component space;

projecting an unknown spectrum onto the principal component space; and

determining the identity of the unknown spectrum by its proximity to clusters of the basis set in the principal component space.

2. The real-time computerized gamma-ray signature identification system of claim 1,

wherein the step of providing the library of predetermined spectra includes calculating signature variations to avoid the difficulty of measuring for signature variations.

3. The real-time computerized gamma-ray signature identification system of claim 1,

wherein the library includes predetermined spectra for various nuclear material types and associated intrinsic and extrinsic variations.

4. The real-time computerized gamma-ray signature identification system of claim 3,

wherein the extrinsic variations include shielding configurations.

5. The real-time computerized gamma-ray signature identification system of claim 1,

wherein the library is a comprehensive library having entries representing substantially more than thirty (e.g. hundreds or thousands) spectral models.

6. The real-time computerized gamma-ray signature identification system of claim 1,

further comprising unitizing the spectral entries in the library for performing PCA.

7. The real-time computerized gamma-ray signature identification system of claim 1,

wherein multiple libraries are provided with each having sub-categorized spectral entries.

8. The real-time computerized gamma-ray signature identification system of claim 1,

wherein performing PCA includes receiving input for the number of principal components to be used for PCA.

9. The real-time computerized gamma-ray signature identification system of claim 1,

wherein the basis set indexes individual spectra of the library in the PC space by means of principal component scores.

10. The real-time computerized gamma-ray signature identification system of claim 1,

wherein the step of determining the identity of the unknown spectrum by its proximity to clusters of the basis set in the principal component space includes the determination of Mahalanobis distances to nearby cluster of models.

11. A real-time computerized gamma-ray signature identification system comprising:

at least one library of predetermined spectra for various nuclear material types and configurations;

computer processor means for performing principal component analysis on the library to transform the library into a corresponding basis set representing the library and indexing each spectrum of the library in principal component space;

computer processor means for determining the identity of the unknown spectrum by projecting an unknown spectrum onto the principal component space and determining its proximity to clusters of the basis set in the principal component space.

12. A real-time computerized gamma-ray signature identification system comprising:

at least one library of predetermined spectra for various nuclear material types and configurations;

a principal component analysis module adapted to perform principal component analysis on the library so as to transform the library into a basis set representing the library and indexing each spectrum of the library in principal component space;

an input module for receiving an unknown spectrum to be identified; and

computer processor means for determining the identity of the unknown spectrum by projecting the unknown spectrum onto the principal component space and determining its proximity to clusters of the basis set in the principal component space.

13. An article of manufacture comprising:

a computer usable medium having computer readable program code means embodied therein for identifying in real-time radioactive materials by their gamma-ray signatures, the computer readable program code means comprising;

at least one computer readable library of predetermined spectra for various nuclear material types;

computer readable program code means for performing principal component analysis on the library to produce a transformed basis set representing the library and indexing each spectrum of the library in principal component space;

computer readable program code means for determining the identity of the unknown spectrum by projecting an unknown spectrum onto the principal component space and determining its proximity to clusters of the basis set in the principal component space.