System and method for model adaptation

• US 10,460,472 B2
• Filed: 07/13/2017
• Issued: 10/29/2019
• Est. Priority Date: 05/19/2017
• Status: Active Grant

• Current Assignee: MvTec Software GmbH (MvTec LLC)
• Original Assignee: Universitã¤T Leipzig
• Inventors: Bottger, Tobias, Ulrich, Markus, Steger, Carsten
• Primary Examiner(s): Rudolph, Vincent
• Assistant Examiner(s): Patel, Pinalben
• Application Number:
  US15/649,326
• Publication Number:
  US 20180336699A1
• Time in Patent Office: 838 Days
• CPC Class Codes: G06F 18/22 Matching criteria, e.g. proximity measures
• CPC Class Codes: G06F 18/28 Determining representative reference patterns, e.g. by averaging or distorting; Generating dictionaries
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/10024 Color image
• CPC Class Codes: G06T 3/02 Affine transformations for image registration G06T3/147; for image mosaicing G06T3/4038
• CPC Class Codes: G06T 7/248 involving reference images or patches
• CPC Class Codes: G06T 7/74 involving reference images or patches
• CPC Class Codes: G06V 10/754 involving a deformation of the sample pattern or of the reference pattern; Elastic matching
• CPC Class Codes: G06V 10/772 Determining representative reference patterns, e.g. averaging or distorting patterns; Generating dictionaries

Abstract

A method for adapting a matching model of an object comprising the step of providing an electronic image of the object; providing a matching model of the object, the matching model consisting of a plurality of points; determining a pose of the object in said electronic image by using a matching approach that uses said matching model; transforming the matching model according to said pose, yielding a transformed model; determining for at least one point of said transformed model a corresponding point in said electronic image; and adapting the matching model according to the at least one determined corresponding point.

Claims

1. A method for adapting a matching model of an object comprising the following steps:

(a) providing an electronic image of the object;

(b) providing a matching model of the object, the matching model comprising a plurality of model levels comprising a plurality of points and optionally a plurality of directions;

(c) determining a pose of the object in said electronic image by using a matching approach that uses said matching model, wherein step (c) includes;

(c1) creating an image pyramid of said electronic image, the image pyramid consisting of a plurality of image levels,

(c2) determining the pose of the object on a first image level of the image pyramid with a lower resolution by using a first model level that corresponds to the first image level,

(c3) refining the pose of the object on a second image level of the image pyramid with a higher resolution by using a second model level that corresponds to the second image level;

(d) transforming the matching model according to said pose, yielding a transformed model;

(e) determining for at least one point of said transformed model a corresponding point in said electronic image; and

(f) adapting the matching model according to the at least one determined corresponding point.

2. The method of claim 1, wherein adapting the matching model in step (f) comprises adapting the position of a plurality of points of the matching model.

3. The method of claim 2, wherein adapting the position of a plurality of points of the matching model comprises setting the position of a plurality of points of the matching model to the position of said corresponding points.

4. The method of claim 2, wherein adapting the position of a plurality of points of the matching model comprises setting the position of a plurality of points of the matching model to a position that is a function of the positions of said points of the matching model and of said corresponding points.

5. The method claim 1, wherein the matching approach in step (c) uses a plurality of points of the matching model to determine said pose of the object.

6. The method of claim 1, wherein the matching model consists of a plurality of points and directions and wherein adapting the matching model in step (f) comprises adapting the position of a plurality of points and directions of the matching model.

7. The method of claim 6, wherein the directions are represented by direction vectors.

8. The method of claim 7, wherein the matching approach in step (c) uses a plurality of points and direction vectors of the matching model and computes the dot product of said direction vectors and the corresponding directions in the image to determine said pose of the object.

9. The method of claim 6, wherein the directions are represented by angles.

10. The method of claim 6, wherein the matching approach in step (c) uses a plurality of points and directions of the matching model to determine said pose of the object.

11. The method of claim 6, wherein adapting the position of a plurality of points and directions of the matching model comprises setting the position and direction of a plurality of points of the matching model to the position and direction of said corresponding points.

12. The method of claim 6, wherein adapting the position of a plurality of points and directions of the matching model comprises setting the position and direction of a plurality of points of the matching model to a position and a direction that are functions of the position and direction of the points of the matching model and of the position and direction of said corresponding points.

13. The method of claim 6, wherein step (e) comprises searching for a corresponding point in said electronic image within a neighborhood of the at least one point of said transformed model, wherein the neighborhood is determined by the direction of said at least one point.

14. The method of claim 1, wherein steps (a)-(f) are performed in an iterative manner with at least two iterations and with the resulting adapted matching model in step (f) acting as the input matching model in step (b) in a next iteration.

15. The method of claim 1, wherein steps (d)-(f) are separately performed for each image level and model level, yielding an adapted matching model.

16. The method of claim 1, wherein steps (d)-(f) are performed on the second image level of the image pyramid with the higher resolution yielding adaptations on the second model level with said higher resolution, and wherein the following additional step is performed:

(g) propagating the adaptations on the second model level with said higher resolution to the first model level with lower resolution, yielding an adapted matching model.

17. The method of claim 1, wherein step (e) comprises searching for a corresponding point in said electronic image within a neighborhood of the at least one point of said transformed model.

18. The method of claim 1, wherein step (f) comprises adapting the matching model according to a subset of the at least one determined corresponding point.

19. The method of claim 1, wherein steps (a)-(f) are performed in an iterative manner with at least two iterations and where step (f) is replaced by:

(f) adapting the matching model according to at least one determined corresponding point in the at least one previous or current iteration, and with the resulting adapted matching model in step (f) acting as the input matching model in step (b) in a next iteration.