Three-dimensional (3D) object recognition system using region of interest geometric features

• US 8,553,989 B1
• Filed: 04/27/2010
• Issued: 10/08/2013
• Est. Priority Date: 04/27/2010
• Status: Expired due to Fees

• Current Assignee: HRL Laboratories LLC (The Boeing Co.), The Boeing Co.
• Original Assignee: HRL Laboratories LLC (The Boeing Co.)
• Inventors: Owechko, Yuri, Medasani, Swarup, Nelson, Jim
• Primary Examiner(s): Patel, Jayesh A
• Assistant Examiner(s): Kholdebarin, Iman K
• Application Number:
  US12/799,618
• Time in Patent Office: 1,260 Days
• Field of Search: 382154-224
• US Class Current: 382/224
• CPC Class Codes: G06V 10/25 Determination of region of interest [ROI] or a volume of interest [VOI]
• CPC Class Codes: G06V 10/422 for representing the structure of the pattern or shape of an object therefor
• CPC Class Codes: G06V 20/64 Three-dimensional objects

Abstract

The present invention relates to a method for three-dimensional (3D) object recognition using region of interest geometric features. The method includes acts of receiving an implicit geometry representation regarding a three-dimensional (3D) object of interest. A region of interest (ROI) is centered on the implicit geometry representation such that there is at least one intersection area between the ROI and the implicit geometry representation. Object shape features are calculated that reflect a location of the ROI with respect to the implicit geometry representation. The object shape features are assembled into a feature vector. A classification confidence value is generated with respect to a particular object classification. Finally, the 3D object of interest is classified as a particular object upon the output of a statistical classifier reaching a predetermined threshold.

Claims

1. A computer implemented method for recognizing an object, comprising an act of causing a computer having a processor to perform acts of:

receiving an implicit geometry representation regarding a three-dimensional (3D) object of interest;

centering a region of interest (ROI) on the implicit geometry representation such that there is at least one intersection area between the ROI and the implicit geometry representation;

calculating object shape features that reflect a relationship of the ROI with respect to the implicit geometry representation;

assembling the object shape features into a feature vector;

generating a classification confidence value with respect to a particular object classification; and

classifying the 3D object of interest as a particular object upon the classification confidence value reaching a predetermined threshold.

2. The method as set forth in claim 1, wherein in the act of centering a ROI, the ROI is a columnar volume centered on the object.

3. The method as set forth in claim 1, further comprising an act of centering a series of expanding ROI volumes on the object of interest, resulting in a set of intersection areas and a corresponding set of object shape features, whereby variations of an object'"'"'s shape with radial distance from its z-axis are represented by using the series of expanding ROI volumes of increasing size.

4. The method as set forth in claim 3, further comprising an act of calculating the object shape features for each ROI to form the set of object shape features and concatenating the set of object shape features into an NM-dimensional feature vector where N is the number of features and M is the number of ROIs.

5. The method as set forth in claim 3, wherein in the act of centering a series of expanding ROI volumes on the object of interest, the ROIs are each centered on a different portion of the object of interest.

6. The method as set forth in claim 1, wherein in the act of receiving an implicit geometry representation, the implicit geometry representation is a function representation of three-dimensional (3D) point cloud data sampled on a regular grid.

7. The method as set forth in claim 1, wherein in the act of centering a ROI on the object of interest, the ROT is formed of a plurality of smaller ROI volumes that cover the object of interest, resulting in a set of intersection areas and a corresponding set of object shape features, whereby variations of an object'"'"'s shape are captured by using the plurality of smaller ROT volumes and by repotting the shape features as histograms in order to preserve rotational invariance.

8. The method as set forth in claim 7, wherein in the act of calculating object shape features, the object shape features are calculated based on a number of contiguous object components enclosed by each ROI, the intersection areas between the object components and each ROI, and a volume fraction of each ROI occupied by the object components.

9. A computer program product for object recognition, the computer program product comprising computer-readable instruction means stored on a non-transitory computer-readable medium that are executable by a computer for causing, the computer to perform operations of:

receiving an implicit geometry representation regarding a three-dimensional (3D) object of interest;

centering a region of interest (ROI) on the implicit geometry representation such that there is at least one intersection area between the ROI and the implicit geometry representation;

calculating object shape features that reflect a location of the ROI with respect to the implicit geometry representation;

assembling the object shape features into a feature vector;

generating a classification confidence value with respect to a particular object classification; and

classifying the 3D object of interest as a particular object upon the classification confidence value reaching a predetermined threshold.

10. The computer program product as set forth in claim 9, wherein in the operation of centering a ROI, the ROT is columnar volume centered on the object.

11. The computer program product as set forth in claim 9, further comprising instruction means for causing the processor to perform an operation of centering a series of expanding ROI volumes on the object of interest, resulting in a set of intersection areas and a corresponding set of object shape features, whereby variations of an object'"'"'s shape with radial distance from its z-axis are represented by using the series of expanding ROI volumes of increasing size.

12. The computer program product as set forth in claim 11, further comprising, instruction means for causing, the processor to perform an operation of calculating the object shape features for each ROI to form the set of object shape features and concatenating the set of object shape features into an NM-dimensional feature vector where N is the number of features and M is the number of ROIs.

13. The computer program product as set forth in claim 11, wherein in the operation of centering a series of expanding ROI volumes on the object of interest, the ROIs are each centered on a different portion of the object of interest.

14. The computer program product as set forth in claim 9, wherein in the operation of receiving an implicit geometry representation, the implicit geometry representation is a population or distance function representation of three-dimensional (3D) point cloud data sampled on a regular grid.

15. The computer program product as set forth in claim 9, wherein in the operation of centering a ROI on the object of interest, the ROI is formed of a plurality of smaller ROI volumes that cover the object of interest, resulting in a set of intersection areas and a corresponding set of object shape features, whereby variations of an object'"'"'s shape are captured by using the plurality of smaller ROI volumes and by reporting the shape features as histograms in order to preserve rotational invariance.

16. The computer program product as set forth in claim 15, wherein in the operation of calculating object shape features, the object shape features are calculated based on a number of contiguous object components enclosed by each ROI, the intersection areas between the object components and each ROI, and a volume fraction of each ROI occupied by the object components.

17. A system for object recognition, the system comprising a memory module and a processor, wherein the memory module is encoded with instruction means that are executable by the processor for causing the processor to perform operations of:

receiving an implicit geometry representation regarding a three-dimensional (3D) object of interest;

centering a region of interest (ROI) on the implicit geometry representation such that there is at least one intersection area between the ROI and the implicit geometry representation;

calculating object shape features that reflect a location of the ROI with respect to the implicit geometry representation;

assembling the object shape features into a feature vector;

generating, a classification confidence value with respect to a particular object classification; and

classifying the 3D object of interest as a particular object upon the classification confidence value reaching a predetermined threshold.

18. The system as set forth in claim 17, wherein in the operation of centering a ROI, the ROI is columnar volume centered on the object.

19. The system as set forth in claim 17, further comprising instruction means for causing the processor to perform an operation of centering a series of expanding ROI volumes on the object of interest, resulting in a set of intersection areas and a corresponding set of object shape features, whereby variations of an object'"'"'s shape with radial distance from its z-axis are represented by using the series of expanding ROI volumes of increasing size.

20. The system as set forth in claim 19, further comprising instruction means for causing the processor to perform an operation of calculating the object shape features for each ROI to form the set of object shape features and concatenating the set of object shape features into an NM-dimensional feature vector where N is the number of features and M is the number of ROIs.

21. The system as set forth in claim 19, wherein in the operation of centering a series of expanding ROI volumes on the object of interest, the ROIs are each centered on a different portion of the object of interest.

22. The system as set forth in claim 17, wherein in the operation of receiving an implicit geometry representation, the implicit geometry representation is a population or distance function representation of three-dimensional (3D) point cloud data sampled on a regular grid.

23. The system as set forth in claim 17, wherein in the operation of centering a ROI on the object of interest, the ROI is formed of a plurality of smaller ROI volumes that cover the object of interest, resulting in a set of intersection areas and a corresponding set of object shape features, whereby variations of an object'"'"'s shape are captured by using the plurality of smaller ROI volumes and by reporting the shape features as histograms in order to preserve rotational invariance.

24. The system as set forth in claim 23, wherein in the operation of calculating object shape features, the object shape features are calculated based on a number of contiguous object components enclosed by each ROI, the intersection areas between the object components and each ROI, and a volume fraction of each ROI occupied by the object components.