X-ray diffraction imaging system with signal aggregation across voxels containing objects and method of operating the same

Patent metadata

Field	Value
Publication number	US-9405990-B2
Application number	US-201414463459-A
Country	US
Kind code	B2
Filing date	Aug 19, 2014
Priority date	Aug 19, 2014
Publication date	Aug 2, 2016
Grant date	Aug 2, 2016

Abstract

A method of performing a security inspection of a container including a plurality of objects includes irradiating the container with polychromatic x-rays, reconstructing a 4-D voxelized representation defining a momentum transfer spectrum, and generating a 3-D image by determining a single value at each voxel as a function of the momentum transfer spectrum. The method further includes segmenting the voxels into segments of contiguous voxels that map onto at least one object. The method also includes computing an aggregated momentum transfer spectrum over at least a portion of the contiguous voxels through at least one of aggregating the spectra of the contiguous voxels for each segment and defining a supervoxel that includes voxels with the object exclusively and voxels in a neighborhood, regardless of association with the object. The method further includes classifying the object as one of a threat and a non-threat based on the aggregated momentum transfer spectrum.

First claim

What is claimed is:

1. A computer-implemented method of performing a security inspection of a container including a plurality of objects therein using a computing device including at least one processor coupled to a memory device, said method comprising:
irradiating the container with polychromatic x-rays;
reconstructing, using the computing device, a four-dimensional (4-D) voxelized representation of a plurality of scatter cross-sections within the container, wherein the 4-D voxelized representation includes a plurality of voxels, a first three dimensions represent a spatial location of each voxel of the plurality of voxels, and a fourth dimension represents a plurality of momentum transfer values defining a momentum transfer spectrum of the container;
generating at least one three-dimensional (3-D) image by determining a single value at each voxel of the plurality of voxels as a function of the momentum transfer spectrum;
segmenting the plurality of voxels into a plurality of segments of contiguous voxels, wherein each segment of the plurality of segments includes a plurality of contiguous voxels, wherein at least a portion of the plurality of segments at least partially maps onto at least one object of the plurality of objects;
computing one or more aggregated momentum transfer spectra over the at least a portion of the plurality of segments by using at least a portion of the plurality of contiguous voxels as guides for aggregation;
classifying the one or more aggregated momentum transfer spectra as one of threat and non-threat; and
distinguishing the at least one object as one of a threat segment and a non-threat segment based on the one or more aggregated momentum transfer spectra.

2. The method in accordance with claim 1 , wherein generating at least one 3-D image comprises summing the plurality of scatter cross-sections over substantially all of the plurality of momentum transfer values within the momentum transfer spectrum.

3. The method in accordance with claim 1 further comprising de-noising the plurality of voxels, thereby facilitating a smoothing of the 3-D image.

4. The method in accordance with claim 1 , wherein segmenting the plurality of voxels into a plurality of segments of contiguous voxels comprises at least one of:
thresholding each voxel of the plurality of voxels comprising comparing the single value of each voxel of the plurality of voxels to a predetermined threshold value; and
performing connected component labeling of each voxel of the plurality of voxels.

5. The method in accordance with claim 4 , wherein segmenting the plurality of voxels into a plurality of segments of contiguous voxels further comprises performing image erosion after thresholding each voxel of the plurality of voxels.

6. The method in accordance with claim 1 , wherein segmenting the plurality of voxels into a plurality of segments of contiguous voxels further comprises splitting at least one segment of the at least a portion of the plurality of segments into a plurality of separated segments of contiguous voxels.

7. The method in accordance with claim 6 , wherein splitting at least one segment of contiguous voxels of the at least a portion of the plurality of segments into a plurality of separated segments of contiguous voxels comprises:
determining an x-axis, a y-axis, and a z-axis that together span 3-D space;
computing an x-coordinate, a y-coordinate, and a z-coordinate for each voxel of the at least one segment;
rotating the at least one segment about the y-axis and the z-axis;
determining a projected sum of intensities, on the x-axis, of the at least one segment through rotation about the y-axis and the z-axis;
determining at least one minima of the projected sum of intensities; and
determining at least one split plane for the at least one minima.

8. The method in accordance with claim 1 , wherein computing one or more aggregated momentum transfer spectra over at least a portion of the plurality of segments comprises aggregating the spectra of the plurality of contiguous voxels for each segment of the plurality of segments, thereby generating an aggregated spectra value for each segment.

9. The method in accordance with claim 1 , wherein computing one or more aggregated momentum transfer spectra over the at least a portion of the plurality of segments by using at least a portion of the plurality of contiguous voxels as guides for aggregation comprises:
defining a neighborhood of voxels comprising:
defining a first voxel within the at least a portion of the plurality of contiguous voxels; and
defining a plurality of second voxels within the at least a portion of the plurality of contiguous voxels, wherein the plurality of second voxels have a determined spatial proximity to the first voxel; and

aggregating momentum transfer spectra from all of the voxels in the neighborhood.

10. The method in accordance with claim 9 , wherein defining a neighborhood of voxels comprises:
determining an x-axis, a y-axis, and a z-axis that together span 3-D space; and
at least one of:
determining the second voxels extending along each of the x-axis, the y-axis, and the z-axis from the first voxel for a predetermined number of voxels; and
determining the second voxels extending along each of the x-axis, the y-axis, and the z-axis from the first voxel for a predetermined distance.

11. An x-ray diffraction imaging (XDI) system comprising:
at least one x-ray source configured to irradiate a container including a plurality of objects therein with polychromatic x-rays;
at least one detector configured to detect scattered x-rays after the polychromatic x-rays have passed through the container; and
a computing device coupled to said at least one detector, said computing device comprising at least one processor and a memory device coupled to said at least one processor, said at least one processor configured to:
reconstruct a four-dimensional (4-D) voxelized representation of a plurality of scatter cross-sections within the container, wherein the 4-D voxelized representation includes a plurality of voxels, a first three dimensions represent a spatial location of each voxel of the plurality of voxels, and a fourth dimension represents a plurality of momentum transfer values defining a momentum transfer spectrum of the container;
generate at least one three-dimensional (3-D) image by determining a single value at each voxel of the plurality of voxels as a function of the momentum transfer spectrum;
segment the plurality of voxels into a plurality of segments of contiguous voxels, wherein each segment of the plurality of segments includes a plurality of contiguous voxels, wherein at least a portion of the plurality of segments at least partially maps onto at least one object of the plurality of objects;
compute one or more aggregated momentum transfer spectra over the at least a portion of the plurality of segments by using the at least a portion of the plurality of contiguous voxels as guides for aggregation;
classify the one or more aggregated momentum transfer spectra as one of threat and non-threat; and
distinguish the at least one object as one of a threat segment and a non-threat segment based on the one or more aggregated momentum transfer spectra.

12. The XDI system in accordance with claim 11 , wherein said at least one processor is further configured to generate at least one 3-D image through summing the plurality of scatter cross-sections over substantially all of the plurality of momentum transfer values within the momentum transfer spectrum.

13. The XDI system in accordance with claim 11 , wherein said at least one processor is furth

Assignees

• Morpho Detection Llc

Inventors

• Skatter Sondre
• Merzbacher Matthew Allen
• Harding Geoffrey
• Basu Samit Kumar
• Zienert Gabriel

Classifications

• G01N23/20083 Primary

  by using a combination of at least two measurements at least one being a transmission measurement and one a scatter measurement · CPC title

• G06T12/20

  Inverse problem, i.e. transformations from projection space into object space · CPC title

• G06F18/24

  Classification techniques · CPC title

• G01N2223/206

  sources operating at different energy levels · CPC title

• G01N2223/639

  material in a container · CPC title