Item classification using localized CT value distribution analysis
Among other things, one or more systems and/or techniques for classifying an item disposed within an object are provided herein. A three-dimensional image of the object (e.g., a bag) is segmented into a set of item representations (e.g., laptop, thermos, etc.). An item is identified from the set of item representations based upon item features of the item, such as the laptop that could be used to conceal an item of interest such as an explosive. A region comprising a three-dimensional image of the item is divided into a set of sub-regions (e.g., a first sub-region encompassing a screen, a second sub-region encompassing a motherboard, etc.). The item is classified as a potential first type of item (e.g., an explosive laptop) when any sub-region has a number of voxels, with computed tomography (CT) values within a range of known CT values for a first type of item, exceeding a threshold.
1. A method for classifying an item disposed within an object, comprising:
receiving a three-dimensional image of the item;
dividing the three-dimensional image of the item into a set of sub-regions;
for a first sub-region of the set of sub-regions:
binning voxels disposed within the first sub-region into a first set of bins based upon computed tomography (CT) values of each voxel disposed within the first sub-region; and
classifying the item as a potential first type of item when a number of voxels, within the first sub-region having CT values within a range of known CT values for a first type of item, exceeds a specified threshold.
2. The method of claim 1 , wherein the first type of item corresponds to a potential explosive item and the range of known CT values corresponds to a range of known CT values for explosives.
3. The method of claim 1 , wherein a CT value corresponds to a density value.
4. The method of claim 1 , wherein a CT value corresponds to a z-effective value.
5. The method of claim 1 , comprising:
receiving a three-dimensional image of the object in which item is disposed; and
segmenting the three-dimensional image of the object to isolate the three-dimensional image of the item from a remaining portion of the three-dimensional image of the object.
6. The method of claim 5 , comprising:
determining eigenvectors of the item using the three-dimensional image of the item; and
defining a region that comprises the three-dimensional image of the item and a set of voxels neighboring the three-dimensional image of the item and not representative of the item.
7. The method of claim 6 , wherein the dividing comprises dividing the region, including the three-dimensional image of the item, into a set of sub-regions.
8. The method of claim 6 , wherein the receiving a three-dimensional image of the item comprises receiving the set of voxels neighboring the three-dimensional image of the item.
9. The method of claim 1 , comprising:
for a second sub-region of the set of sub-regions:
binning voxels disposed within the second sub-region into the first set of bins based upon CT values of each voxel disposed within the second sub-region; and
classifying the item as the potential first type of item when a number of voxels, within the second sub-region that have CT values within the range of known CT values for the first type of item, exceeds the specified threshold.
10. The method of claim 9 , wherein the item is classified as the potential first type of item when the number of voxels within the second sub-region that have CT values within the range of known CT values for the first type of item exceeds the specified threshold irrespective of whether the number of voxels within the first sub-region having CT values within the range of known CT values for the first type of item exceeds the specified threshold.
11. The method of claim 9 , wherein the first sub-region and the second sub-region partially overlap.
12. The method of claim 9 , wherein the first sub-region and the second sub-region are mutually exclusive.
13. The method of claim 9 , wherein the first sub-region has a first thickness and the second sub-region has a second thickness different than the first thickness.
14. The method of claim 1 , comprising:
determining eigenvectors of the item using the three-dimensional image of the item, wherein each sub-region of the set of sub-regions corresponds to a slice of the item along a first Eigen vector.
15. The method of claim 1 , comprising:
computing a histogram for the first sub-region based upon the CT values of each voxel disposed within the first sub-region.
16. A computing device comprising:
a processing unit; and
memory comprising processor-executable instructions that when executed by the processing unit;
cause performance of operations, the operations comprising:
segmenting a three-dimensional image of an object into a set of item representations;
calculating item features of the set of item representations;
identifying a potential concealment item from the set of item representations based upon item features of a potential concealment item representation, of the set of item representations, for the potential concealment item;
defining a region within the three-dimensional image of the object that encompasses the potential concealment item;
dividing the region into a set of sub-regions; and
for a first sub-region of the set of sub-regions, classifying the potential concealment item as comprising a potential first type of item when a number of voxels, within the first sub-region having CT values within a range of known CT values for a first type of item, exceeds a specified threshold.
17. The computing device of claim 16 , wherein an item feature is indicative of an electronics device.
18. The computing device of claim 16 ,
for a second sub-region of the set of sub-regions, classifying the potential concealment item as comprising the potential first type of item when a number of voxels, within the second sub-region and having CT values within the range of known CT values for the first type of item, exceeds the specified threshold.
19. The computing device of claim 18 , wherein the potential concealment item is classified as comprising the potential first type of item within the first sub-region but not within the second sub-region.
20. The computing device of claim 18 , wherein the potential concealment item is classified as comprising a first portion of the potential first type of item within the first sub-region and a second portion of the potential first type of item within the second sub-region.
21. A non-transitory computer readable medium comprising computer executable instructions that when executed via a processing unit cause performance of operations, the operations comprising:
segmenting a three-dimensional image of an object into a set of item representations;
calculating item features of the set of item representations;
identifying a potential concealment item from the set of item representations based upon item features of a potential concealment item representation, of the set of item representations, for the potential concealment item;
defining a region within the three-dimensional image of the object that encompasses the potential concealment item;
dividing the region into a first sub-region, a second sub-region, and a third sub-region;
evaluating CT values of voxels within the first sub-region, the second sub-region, and the third sub-region to determine that an instance of a first type of object exists within any of the first sub-region, the second sub-region, or the third sub-region; and
classifying the potential concealment item as comprising a potential first type of item when the first type of object is determined to exist within any of the first sub-region, the second sub-region, or the third sub-region.