3D beverage container localizer

• US 8,116,519 B2
• Filed: 06/16/2008
• Issued: 02/14/2012
• Est. Priority Date: 09/26/2007
• Status: Active Grant

• Current Assignee: Honda Motor Co., Ltd. (Honda Motor Company)
• Original Assignee: Honda Motor Co., Ltd. (Honda Motor Company)
• Inventors: Ng-Thow-Hing, Victor, Wu, Qingquan
• Primary Examiner(s): Johns, Andrew W
• Assistant Examiner(s): GORADIA, SHEFALI DINESH
• Application Number:
  US12/140,126
• Publication Number:
  US 20090080699A1
• Time in Patent Office: 1,338 Days
• Field of Search: 382/100, 382/103
• US Class Current: 382/103
• CPC Class Codes: G06T 7/73 using feature-based methods
• CPC Class Codes: G06V 10/48 by mapping characteristic values of the pattern into a parameter space, e.g. Hough transformation
• CPC Class Codes: G06V 20/64 Three-dimensional objects

Abstract

Objects placed on a flat surface are identified and localized by using a single view image. The single view image in the perspective projection is transformed to a normalized image in a pseudo plan to view to enhance detection of the bottom or top shapes of the objects. One or more geometric features are detected from the normalized image by processing the normalized image. The detected geometric features are analyzed to determine the identity and the location the objects on the flat surface.

Claims

1. A method of detecting and localizing an object, comprising:

receiving a first single view image including the object, the first single view image in a perspective projection;

mapping the single view image to a normalized image by a homography transform, the normalize image representing a pseudo plan view of the object;

detecting first geometric features from the normalized image;

analyzing the detected first geometric features to identify the object;

generating first object information including the identity of the object; and

outputting the first object information.

2. The method of claim 1, further comprising:

generating location information of the object by analyzing the detected first geometric features, the location information representing location of the object; and

adding the location information to the first object information.

3. The method of claim 1, wherein detecting the first geometric features comprises:

detecting edges in the normalized image; and

detecting circles in the normalized image by performing Hough circle transform on the detected edges.

4. The method of claim 1, further comprising:

detecting second geometric features from the normalized image; and

selecting one or more first geometric features as outlining the object responsive to identifying that the first geometric features intersect with two or more second geometric features.

5. The method of claim 4, wherein detecting the second geometric features comprises:

detecting edges in the normalized image; and

detecting straight lines in the normalized image by performing Hough line transform on the detected edges.

6. The method of claim 4, further comprising:

determining the object as a first object responsive to determining that a distance between two second geometric features that intersect with the first geometric features is smaller than a threshold; and

determining the object as a second object responsive to determining that the distance is not smaller than the threshold.

7. The method of claim 1, further comprising determining the first object information as incorrect responsive to difference between the first object information and second object information exceeding a threshold, the second location information determined from a second single view image preceding the first single view image.

8. The method of claim 1, further comprising determining location of the object based on the two dimensional location of the first feature and location information of another object in the normalized image.

9. A computer readable storage medium structured to store instructions adapted to detect and localize an object, the instructions when executed, cause the processor to:

receive a first single view image including the object, the first single view image in a perspective projection;

map the single view image to a normalized image by a homography transform, the normalize image representing a pseudo plan view of the object;

detect first geometric features from the normalized image;

analyze the detected first geometric features to identify the object;

generate first object information including the identity of the object; and

output the first object information.

10. The computer readable storage medium of claim 9, further comprising instructions to:

generate location information of the object by analyzing the detected first geometric features, the location information representing location of the object; and

add the location information to the first object information.

11. The computer readable storage medium of claim 9, wherein instructions to detect the first geometric features comprise instructions to:

detect edges in the normalized image; and

detect circles in the normalized image by performing Hough circle transform on the detected edges.

12. The computer readable storage medium of claim 9, further comprising instructions to:

detect second geometric features from the normalized image; and

select one or more first geometric features as outlining the object responsive to identifying that the first geometric features intersect with two or more second geometric features.

13. The computer readable storage medium of claim 12, wherein the instructions to detect the second geometric features comprise instructions to:

detect edges in the normalized image; and

detect line features in the homography transformed image by performing Hough line transform on the detected edges.

14. The computer readable storage medium of claim 12, further comprising instructions to:

determine the object as a first object responsive to determining that a distance between two second geometric features that intersect with the first geometric features is smaller than a threshold; and

determine the object as a second object responsive to determining that the distance is not smaller than the threshold.

15. The computer readable storage medium of claim 9, further comprising instructions to determine the first object information as incorrect responsive to difference between the first object information and second object information exceeding a threshold, the second location information determined from a second single view image preceding the first single view image.

16. The computer readable storage medium of claim 9, further comprising instructions to determine location of the object based on the two dimensional location of the first feature and location information of another object in the normalized image.

17. A system for localizing and detecting an object, comprising:

a pre-processing module configured to receive a first single view image including the object, the first single view image in a perspective projection, the pre-processing module configured to map the single view image to a normalized image by a homography transform, the normalize image representing a pseudo plan view of the object;

a feature extraction module configured to detect first geometric features in the normalized image; and

an object information generator configured to detect first geometric features from the normalized image and analyzing the detected first geometric features to identify the object, the object information generator generating first object information including the identity of the object.

18. The system of claim 17, wherein the object information generator is further configured to generate location information of the object by analyzing the detected first geometric features, the location information representing location of the object, the object information generator including the location information in the first object information.

19. The system of claim 17, wherein the feature extraction module is configured to detect the first geometric features by detecting edges in the normalized image, and to detect circle features in the normalized image by performing Hough circle transform on the detected edges.

20. The system of claim 17, wherein the feature extraction module is configured to detect second geometric features from the normalized image and select one or more first geometric features as representing the object by identifying whether the first geometric features intersect with two or more second geometric features.

21. The system of claim 20, wherein the feature extraction module is configured to detect the second geometric features by detecting edges in the normalized image and detecting line features in the normalized image by performing Hough line transform on the detected edges.

22. The system of claim 17, further comprising an object information buffer configured to store second object information determined from a second single view image preceding the first single view image, the object information generator configured to determine the first object information as incorrect responsive to a difference between the first object information and the second object information exceeding a threshold.