Detecting concealed threats
View Patent ↗Potential threat items may be concealed inside objects, such as portable electronic devices, that are subject to imaging for example, at a security checkpoint. Data from an imaged object can be compared to pre-determined object data to determine a class for the imaged object. Further, an object can be identified inside a container (e.g., a laptop inside luggage). One-dimensional Eigen projections can be used to partition the imaged object into partitions, and feature vectors from the partitions and the object image data can be used to generate layout feature vectors. One or more layout feature vectors can be compared to training data for threat versus non-threat-containing items from the imaged object's class to determine if the imaged object contains a potential threat item.
1. A system for identifying a potentially threat-containing object that has been subjected to imaging using an imaging apparatus, comprising:
a processor; and
memory comprising instructions that when executed via the processor implement at least some of:
an object partitioner configured to partition an image of an object into two or more object partitions by:
projecting the image along a first Eigen vector, associated with a first Eigen value, to generate an Eigen projection;
projecting the Eigen projection along a second Eigen vector, associated with a second Eigen value, to generate a one-dimensional projection, the second Eigen value less than the first Eigen value; and
identifying peak modes in the one-dimensional projection, a first peak mode corresponding to a first object partition of the two or more object partitions and a second peak mode corresponding to a second object partition of the two or more object partitions;
a layout vector generator configured to generate a layout feature vector corresponding to the first object partition based at least in part upon one or more features associated with the first object partition; and
a threat determiner configured to compare the layout feature vector to one or more predetermined boundaries to determine whether the object is a potentially threat-containing object.
2. The system of claim 1 , comprising an object segmentor configured to segment the image of the object from an image of a container comprising the object.
3. The system of claim 2 , the object segmentor comprising:
a one dimensional image data processor configured to generate a one dimensional representation of an axis of the image of the container;
an axes data processor configured to compare data points in the one dimensional representation of the axis with a predetermined threshold;
a continuation score generator configured to generate a continuation score corresponding to a number of consecutive data points greater than the predetermined threshold; and
a continuation score comparison component configured to compare the continuation score with a predetermined score to determine whether the number of consecutive data points represent an object of a class of objects where partitioning is desirable.
4. The system of claim 1 , the object partitioner comprising an Eigen vector determiner configured to determine Eigen vectors for the object including the first Eigen vector and the second Eigen vector.
5. The system of claim 1 , comprising a sub-region extractor configured to identify a sub-region in the first object partition, the sub-region extractor comprising:
a sub-region identifier configured to identify a potential sub-region in the first object partition; and
a sub-region comparison component configured to compare the potential sub-region with predetermined sub-region data to identify whether the potential sub-region belongs to a class of known sub-regions.
6. The system of claim 1 , the threat determiner configured to compare the layout feature vector to a discriminating boundary corresponding to a class of objects to determine whether the object is associated with the class of objects.
7. A method for identifying a potentially threat-containing object that has been subjected to imaging using an imaging apparatus, comprising:
partitioning an image of an object into two or more object partitions, comprising:
projecting the image along a first Eigen vector, associated with a first Eigen value, to generate an Eigen projection;
projecting the Eigen projection along a second Eigen vector, associated with a second Eigen value, to generate a one-dimensional projection, the second Eigen value less than the first Eigen value; and
identifying peak modes in the one-dimensional projection, a first peak mode corresponding to a first object partition of the two or more object partitions and a second peak mode corresponding to a second object partition of the two or more object partitions;
generating a layout feature vector corresponding to the first object partition based at least in part upon one or more features associated with the first object partition; and
determining whether the object is a potentially threat-containing object comprising comparing the layout feature vector to one or more predetermined boundaries for potential threat objects.
8. The method of claim 7 , comprising:
segmenting the image of the object from an image of a container comprising the object.
9. The method of claim 7 , comprising:
determining whether the object belongs to a class of known objects; and
performing the acts of partitioning, generating, and determining whether the object is a potentially threat-containing object merely when the object belongs to the class of known objects.
10. The method of claim 9 , the determining whether the object belongs to a class of known objects comprising:
performing an Eigen analysis on the image of the object to determine Eigen vectors for the object including the first Eigen vector and the second Eigen vector; and
comparing features of the object to features of one or more known objects comprised in the class of known objects based at least in part upon the Eigen vectors to determine whether the object belongs to the class of known objects.
11. The method of claim 7 , comprising identifying the object, comprising:
generating a one dimensional representation of an axis of the image of the object; and
comparing data points in the one dimensional representation of the axis with a predetermined threshold.
12. The method of claim 11 , the identifying comprising:
generating a continuation score corresponding to a number of consecutive data points greater than a predetermined threshold; and
comparing the continuation score with a predetermined score to determine whether the object belongs to a class of known objects, wherein
the acts of partitioning, generating, and determining whether the object is a potentially threat-containing object are not performed when the object does not belong to the known class of objects.
13. The method of claim 7 , the partitioning comprising:
performing an Eigen analysis on the image of the object to determine Eigen vectors for the object including the first Eigen vector and the second Eigen vector.
14. A non-transitory computer readable medium comprising processor-executable instructions that when executed perform a method for identifying a potentially threat-containing object that has been subjected to imaging using an imaging apparatus, the method comprising:
partitioning an image of an object into two or more object partitions, comprising:
projecting the image along a first Eigen vector, associated with a first Eigen value, to generate an Eigen projection;
projecting the Eigen projection along a second Eigen vector, associated with a second Eigen value, to generate a one-dimensional projection, the second Eigen value less than the first Eigen value; and
identifying peak modes in the one-dimensional projection, a first peak mode corresponding to a first object partition of the two or more object partitions and a second peak mode corresponding to a second object partition of the two or more object partitions;
generating a layout feature vector corresponding to the first object partition based at least in part upon one or more features associated with the first object partition; and
determining whether the object is a potentially threat-containing object comprising comparing the layout feature vector to one or more predetermined boundaries for potential threat objects.
15. The non-transitory computer readable medium of claim 14 , the method comprising:
segmenting the image of the object from an image of a container comprising the object.
16. The non-transitory computer readable medium of claim 14 , the method comprising:
determining whether the object belongs to a class of known objects; and
performing the acts of partitioning, generating, and determining whether the object is a potentially threat-containing object merely when the object belongs to the class of known objects.
17. The non-transitory computer readable medium of claim 16 , the determining whether the object belongs to a class of known objects comprising:
performing an Eigen analysis on the image of the object to determine Eigen vectors for the object including the first Eigen vector and the second Eigen vector; and
comparing features of the object to features of one or more known objects comprised in the class of known objects based at least in part upon the Eigen vectors to determine whether the object belongs to the class of known objects.
18. The non-transitory computer readable medium of claim 14 , the method comprising:
identifying a potential sub-region in the first object partition; and
comparing the potential sub-region with predetermined sub-region data to identify whether the potential sub-region belongs to a class of known sub-regions.
19. The non-transitory computer readable medium of claim 14 , the object comprising a laptop.
20. The non-transitory computer readable medium of claim 14 , wherein the first object partition corresponds to a first distinct layer of the object and the second object partition corresponds to a second distinct layer of the object, the first distinct layer different than the second distinct layer.