Compound object separation
View Patent ↗Representations of an object can comprise two or more separate sub-objects, producing a compound object. Compound objects can affect the quality of object visualization and threat identification. As provided herein, a compound object can be separated into sub-objects based on object morphological properties (e.g., an object's shape, surface area). Further, a potential compound object can be split into sub-objects, for example, eroding one or more outer layers of volume space (e.g., voxels) from the potential compound object. Additionally, a volume of a representation of the sub-objects in an image can be reconstructed, for example, by generating sub-objects that have a combined volume approximate to that of the compound object. Furthermore, sub-objects, which can be parts of a same physical object, but may have been erroneously split, can be identified and merged using connectivity and compactness based techniques.
1. A system for compound object separation, comprising:
a compound object splitter component configured to generate, from a representation of a potential compound object, a first representation of a first potential sub-object and a second representation of a second potential sub-object by eroding a first set of voxels from the representation of the potential compound object; and
a volume reconstructor component configured to generate a first image representing the first potential sub-object from the first representation and a second image representing the second potential sub-object from the second representation, the volume reconstructor component comprising:
a neighboring eroded voxel identifier component configured to group one or more voxels, of the first set of voxels, that are spatially connected to the first representation to generate a second set of voxels; and
an eroded voxel dissolver component configured to dissolve the second set of voxels to generate at least a first portion of the first image.
2. The system of claim 1 , comprising a connectivity merger component configured to merge the first image with the second image when a connectivity ratio for the first image and the second image is within a pre-determined connectivity threshold.
3. The system of claim 2 , the connectivity ratio substantially corresponding to a number of pairs of spatially connected voxels between the first image and the second image divided by a combined volume of the first potential sub-object as represented by the first image and the second potential sub-object as represented by the second image.
4. The system of claim 1 , comprising a compactness merger component configured to merge the first image with the second image when a compactness ratio for the first image and the second image is within a pre-determined compactness threshold.
5. The system of claim 4 , the compactness merger component comprising:
an Eigen-box generator component configured to generate a first sub-object Eigen-box based upon the first image, a second sub-object Eigen-box based upon the second image, and a union Eigen-box for a union of the first potential sub-object and the second potential sub-object based upon the first image and the second image;
a fill ratio determiner component configured to calculate a first fill ratio for the first sub-object Eigen box, a second fill ratio for the second sub-object Eigen box, and a third fill ratio for the union Eigen-box; and
a compactness ratio generator component configured to compare the first fill ratio, the second fill ratio, and the third fill ratio.
6. The system of claim 1 , comprising an entry control component configured to identify the representation of the potential compound object based upon one or more features of the potential compound object.
7. The system of claim 1 , the compound object splitter component comprising a volume space eroder component configured to erode the first set of voxels from the representation of the potential compound object to generate an eroded representation of the potential compound object.
8. The system of claim 7 , comprising a sub-object identifier component configured to identify the first representation and the second representation from the eroded representation of the potential compound object.
9. The system of claim 1 , the volume reconstructor component comprising a dilated voxel dissolver component configured to dissolve a third set of voxels to generate at least a second portion of the first image.
10. The system of claim 9 , the volume reconstructor component comprising:
a neighboring dilated voxel identifier component configured to identify one or more voxels that have been dilated and that are spatially connected to the first representation to generate the third set of voxels.
11. A method for compound object separation, comprising:
eroding a first set of voxels from a representation of a potential compound object to generate a first representation of a first potential sub-object and a second representation of a second potential sub-object;
generating a first image representing the first potential sub-object from the first representation;
generating a second image representing the second potential sub-object from the second representation;
generating a first sub-object Eigen-box based upon the first image;
generating a second sub-object Eigen-box based upon the second image;
generating a union Eigen-box for a union of the first potential sub-object and the second potential sub-object based upon the first image and the second image;
determining a first fill ratio for the first sub-object Eigen-box, a second fill ratio for the second sub-object Eigen box, and a third fill ratio for the union Eigen-box;
determining a compactness ratio describing a compactness between the first potential sub-object and the second potential sub-object based upon a comparison of the first fill ratio, the second fill ratio, and the third fill ratio; and
merging the first image with the second image when the compactness ratio meets a pre-determined compactness threshold.
12. The method of claim 11 , the merging comprising merging the first image with the second image when a connectivity ratio for the first image and the second image meets a pre-determined connectivity threshold.
13. The method of claim 12 , the connectivity ratio substantially corresponding to a number of pairs of spatially connected voxels between the first image and the second image divided by a combined volume of the first potential sub-object as represented by the first image and the second potential sub-object as represented by the second image.
14. The method of claim 11 , the determining a first fill ratio comprising comparing a first volume of the first sub-object Eigen-box that is filled by the first potential sub-object to a second volume of the first sub-object Eigen-box that is not filled by the first potential sub-object.
15. The method of claim 11 , comprising, before the eroding, identifying the representation of the potential compound object, the identifying comprising comparing one or more object feature values to one or more corresponding object feature threshold values.
16. The method of claim 11 , the generating a first image comprising at least one of:
dissolving an eroded voxel; or
dissolving a dilated voxel.
17. A system for compound object separation, comprising:
a compound object splitter component configured to generate, from a representation of a potential compound object, a first representation of a first potential sub-object and a second representation of a second potential sub-object by eroding a first set of voxels from the representation of the potential compound object; and
a volume reconstructor component configured to generate a first image representing the first potential sub-object from the first representation and a second image representing the second potential sub-object from the second representation, the volume reconstructor component comprising:
a neighboring dilated voxel identifier component configured to identify one or more voxels that have been dilated and that are spatially connected to the first representation to generate a second set of voxels; and
a dilated voxel dissolver component configured to dissolve the second set of voxels to generate at least a first portion of the first image.
18. The system of claim 17 , the volume reconstructor component comprising an eroded voxel dissolver component configured to dissolve at least one voxel of the first set to generate at least a second portion of the first image.
19. The system of claim 17 , comprising a compactness merger component configured to merge the first image with the second image when a compactness ratio for the first image and the second image is within a pre-determined compactness threshold.
20. The system of claim 19 , the compactness merger component comprising:
an Eigen-box generator component configured to generate a first sub-object Eigen-box based upon the first image, a second sub-object Eigen-box based upon the second image, and a union Eigen-box for a union of the first potential sub-object and the second potential sub-object based upon the first image and the second image;
a fill ratio determiner component configured to calculate a first fill ratio for the first sub-object Eigen box, a second fill ratio for the second sub-object Eigen box, and a third fill ratio for the union Eigen-box; and
a compactness ratio generator component configured to compare the first fill ratio, the second fill ratio, and the third fill ratio.