Method of registration in a contraband detection system
View Patent ↗A multi-level contraband detection system. At a first level, the system obtains volumetric information about an item under inspection. The volumetric information provides a basis for identifying suspect objects in the item under inspection and their locations. The location information is expressed in a first coordinate system relative to the device used for first level scanning. When first level scanning identifies a suspicious object, the item under inspection is passed to a second level scanner that can take further measurements on the suspicious objects. The second level machine is controlled in a second coordinate system. A translation between the two coordinate systems is obtained by registering a multi-dimensional image obtained at the first level with positioning information obtained at the second level. Registration is performed using a coarse and then a fine registration process for quick and accurate registration.
1. A method of registering a first image with a second image, the method comprising;
a) identifying a plurality of points in the first image;
b) identifying a plurality of points in the second image;
c) for each of a plurality of pairs of subsets, each pair of subsets comprising a subset of the plurality points in the first image and a subset of the plurality of points in the second image, computing a transformation that maps the subset of points in the first image to the subset of points in the second image;
d) identifying a likely transformation from the transformations for each of the plurality of subsets; and
e) using the likely transformation in registering the first image and the second image.
2. The method of claim 1 , wherein identifying a likely transformation comprises selecting a transformation based on the frequency of occurrence of the likely transformation in the transformations computed for the plurality of pairs of subsets.
3. The method of claim 1 , wherein identifying a likely transformation comprises deemphasizing transformations for pairs of the plurality of pairs of subsets having a high error.
4. The method of claim 3 , wherein deemphasizing transformations comprises:
i) computing errors in transformation for each of the plurality of pairs;
ii) selecting those transformations having an error below a threshold; and
iii) identifying a likely transformation from the selected transformations.
5. The method of claim 4 , wherein identifying a likely transformation comprises selecting a transformation from a cluster.
6. The method of claim 1 , wherein identifying a plurality of points in the first image comprises extracting features from the first image and selecting the plurality of points on the features.
7. The method of claim 6 , wherein identifying a plurality of points in the second image comprises extracting features from the second image and selecting the plurality of points on the features.
8. The method of claim 6 , wherein computing a transformation that maps the subset of the plurality of points in the first image to the subset of the plurality of points in the second image, comprises mapping a subset consisting essentially of two points in the first image to a subset consisting essentially of two points in the second image.
9. The method of claim 1 , wherein the first image is a three-dimensional image end the second image is a two-dimensional image and identifying a plurality of points in the first image comprises selecting a plurality of locations in the three-dimensional image and projecting those locations into points in a two-dimensional projection.
10. The method of claim 1 , additionally comprising scaling at least one of the first or second images to match the scale of the first image to the second image.
11. A method of operating a contraband detection system to inspect an item, the method comprising:
a) forming a volumetric image of the item in a first coordinate system;
b) identifying a region of interest in the item using the volumetric image;
c) forming a projection image of the item in a second coordinate system;
d) registering the volumetric image to the projection image to identify the region of interest in the second coordinate system; and
e) selecting a region of the item for further inspection based on the region of interest in the second coordinate system.
12. The method of operating a contraband detection system of claim 11 , further comprising probing the selected region of interest in the item using coherent x-ray scattering.
13. The method of operating a contraband detection system of claim 12 , wherein forming a volumetric image comprises processing data obtained from a helical CT scan of the item.
14. The method of operating a contraband detection system of claim 13 , wherein the volumetric image is formed in a first location and the projection image is formed in a second location, and the method further comprises moving the item from the first location to the second location.
15. The method of operating a contraband detection system of claim 14 , wherein moving the item from the first location to the second location comprises moving the item on a plurality of conveyors.
16. A method of operating a contraband detection system to inspect an item, the method comprising:
a) forming a first image of the item in a first coordinate system, the first image including a region of interest;
b) extracting features from the first image to form a first symbolic image;
c) forming a second image of the item in a second coordinate system;
d) extracting features from the second image to form a second symbolic image; and
e) registering the first symbolic image to the second symbolic image to identify the region of interest in the second coordinate system.
17. The method of claim 16 , wherein forming the second image of the item comprises forming a projection image of the item.
18. The method of claim 17 , wherein forming a projection image of the item comprises forming an x-ray projection image of the item.
19. The method of claim 17 , wherein forming the first image of the item comprises forming a volumetric image.
20. The method of claim 16 , further comprising inspecting the region of interest with a coherent x-ray scattering probe.