System and method for automatic selection of 3D alignment algorithms in a vision system

Patent metadata

Field	Value
Publication number	US-10380767-B2
Application number	US-201715634003-A
Country	US
Kind code	B2
Filing date	Jun 27, 2017
Priority date	Aug 1, 2016
Publication date	Aug 13, 2019
Grant date	Aug 13, 2019

Abstract

A system and method for selecting among 3D alignment algorithms in a 3D vision system is provided. The system and method includes a 3D camera assembly to acquire at least a runtime image defined by a 3D point cloud or runtime 3D range image having features of a runtime object and a vision system processor. A training image is provided. It is defined by a 3D point cloud or 3D range image having features of a model. A selection process is operated by the vision processor. It analyzes at least one training region of the training image having the features of the model and determines a distribution of surface normals in the at least one training region. It also selects, based upon a characteristic of the distribution, at least one 3D alignment algorithm from a plurality of available 3D alignment algorithms to align the features of the model with respect to the features of the runtime object.

First claim

What is claimed is:

1. A system for selecting among 3D alignment algorithms in a 3D vision system comprising:
a 3D camera assembly to acquire at least a runtime image defined by a 3D point cloud or runtime 3D range image having features of a runtime object and a vision system processor and a training image defined by a 3D point cloud or 3D range image having features of a model; and
a selection process, operated by the vision processor, that (a) analyzes at least one training region of the training image having the features of the model and determines a distribution of surface normals in the at least one training region, and (b) selects, based upon a characteristic of the distribution, at least one 3D alignment algorithm from a plurality of available 3D alignment algorithms to align the features of the model with respect to the features of the runtime object.

2. The system as set forth in claim 1 wherein the selection process is arranged to locate the at least one training region in the training image upon which to perform training.

3. The system as set forth in claim 1 wherein the training image is acquired by the 3D camera assembly or provided as a synthetic image.

4. The system as set forth in claim 1 wherein the at least one 3D alignment algorithm aligns the features of the model with respect to the features of the runtime object in at least one of a coarse 3D alignment process and a fine 3D alignment process.

5. The system as set forth in claim 1 wherein the characteristic of the distribution is a degree of variance relative to a unimodal distribution of the surface normals, and the selection process is arranged to compare the degree of variance to a threshold.

6. The system as set forth in claim 5 wherein (a) if the variance is higher than a high threshold value, then the selection process is arranged to select an ICP algorithm, and (b) if the variance is lower than a low threshold, then the selection process is arranged to select a hybrid edge-based and ICP algorithm.

7. The system as set forth in claim 6 wherein at least one of the high threshold value and the low threshold value is set by at least one of (a) an automated process or (b) a user-specified process.

8. The system as set forth in claim 7 wherein at least one of the automated process and the user-specified process is based upon a type of object in the training image.

9. The system as set forth in claim 6 wherein, if the variance is between the high threshold and the low threshold, the selection process is arranged to prompt a user to select the at least one 3D alignment algorithm.

10. The system as set forth in claim 1 wherein the 3D camera assembly comprises a plurality of discrete 3D cameras located at spaced-apart positions to image a scene containing the runtime object.

11. A method for selecting among 3D alignment algorithms in a 3D vision system comprising the steps of:
acquiring, with a 3D camera assembly, at least a runtime image defined by a 3D point cloud or runtime 3D range image having features of a runtime object and a vision system processor and a training image defined by a 3D point cloud or 3D range image having features of a model; and
selecting, with the vision processor that (a) analyzes at least one training region of the training image having the features of the model and determines a distribution of surface normals in the at least one training region, and (b) selects, based upon a characteristic of the distribution, at least one 3D alignment algorithm from a plurality of available 3D alignment algorithms to align the features of the model with respect to the features of the runtime object.

12. The method as set forth in claim 11 wherein the step of selecting locates the at least one training region in the training image upon which to perform training.

13. The method as set forth in claim 11 wherein the training image is either acquired by the 3D camera assembly or provided as a synthetic image.

14. The method as set forth in claim 11 further comprising aligning, with the at least one 3D alignment algorithm, the features of the model with respect to the features of the runtime object in at least one of a coarse 3D alignment process and a fine 3D alignment process.

15. The method as set forth in claim 11 wherein the characteristic of the distribution is a degree of variance relative to a unimodal distribution of the surface normals, and the step of selecting includes comparing the degree of variance to a threshold.

16. The method as set forth in claim 15 wherein (a) if the variance is higher than a high threshold value, then the step of selecting selects an ICP algorithm, and (b) if the variance is lower than a low threshold, then the step of selecting selects a hybrid edge-based and ICP algorithm.

17. The method as set forth in claim 16 , further comprising setting at least one of the high threshold value and the low threshold value is set by at least one of (a) an automated process or (b) a user-specified process.

18. The method as set forth in claim 17 wherein at least one of the automated process and the user-specified process is based upon a type of object in the training image.

19. The method as set forth in claim 16 wherein, if the variance is between the high threshold and the low threshold, the step of selecting prompts a user to select the at least one 3D alignment algorithm.

20. The method as set forth in claim 11 wherein the 3D camera assembly comprises a plurality of discrete 3D cameras located at spaced-apart positions to image a scene containing the runtime object.

Assignees

• Cognex Corp

Inventors

• Barker Simon
• Michael David J

Classifications

• G06T7/35

  using statistical methods · CPC title

• G06T2207/10028

  Range image; Depth image; 3D point clouds · CPC title

• G06T7/11

  Region-based segmentation · CPC title

• H04N13/271

  wherein the generated image signals comprise depth maps or disparity maps · CPC title

• H04N13/232

  using fly-eye lenses, e.g. arrangements of circular lenses · CPC title