System and method for producing synthetic golden template image for vision system inspection of multi-layer patterns

• US 8,971,663 B2
• Filed: 09/24/2012
• Issued: 03/03/2015
• Est. Priority Date: 05/21/2012
• Status: Active Grant

• Current Assignee: Cognex Corporation
• Original Assignee: Cognex Corporation
• Inventors: Jacobson, Lowell D., Liu, Gang
• Primary Examiner(s): Alavi, Amir
• Application Number:
  US13/625,021
• Publication Number:
  US 20130308875A1
• Time in Patent Office: 890 Days
• Field of Search: 382/150, 382/154, 382/173, 382/203, 382/209, 382/284
• US Class Current: 382/284
• CPC Class Codes: G06F 18/28 Determining representative reference patterns, e.g. by averaging or distorting; Generating dictionaries
• CPC Class Codes: G06T 2207/10024 Color image
• CPC Class Codes: G06T 2207/20221 Image fusion; Image merging
• CPC Class Codes: G06T 2207/30144 Printing quality
• CPC Class Codes: G06T 7/001 using an image reference approach
• CPC Class Codes: G06V 10/772 Determining representative reference patterns, e.g. averaging or distorting patterns; Generating dictionaries

Abstract

A system and method for generating golden template images in a vision system to inspect an acquired runtime image of an object with a multi-layer printed pattern is provided. The system and method performs a registration process on runtime images using registration models each trained on respective canonical layer mask images, and outputting poses. Based upon the poses, warped layer masks are generated. Combination masks are computed based upon differing combinations of the warped layer masks. Intensity values for pixels of the foreground regions for the combination masks are estimated. The estimated intensity values are then blended associated with the combination masks to generate a golden template image. This golden template image can be used to compare with a runtime image. An exemplary application of this system and method is in print inspection on flat and non-flat surfaces.

Claims

1. A method for generating golden template images in a vision system to inspect an acquired runtime image of an object with a multi-layer printed pattern comprising the steps of:

performing a registration process on runtime images using registration models each trained on respective canonical layer masks, and outputting poses;

based upon the poses, generating warped layer masks;

computing combination masks based upon differing combinations of the warped layer masks;

estimating intensity of regions for the combination masks; and

blending the estimated intensity values associated with the combination masks to generate a golden template image.

2. The method as set forth in claim 1 wherein the step of estimating intensity includes estimating the intensity of an overall background region.

3. The method as set forth in claim 1 wherein the warped layer masks define a fuzzy edge that, in turn, provide combination masks with fuzzy edges in the step of computing.

4. The method as set forth in claim 1 wherein the warped layer masks are generated based upon distortion within a training image.

5. The method as set forth in claim 1 wherein the warped layer masks are generated based upon distortion in the runtime image currently acquired.

6. The method as set forth in claim 1 further comprising, providing pre-warped distortion information to warp the canonical layer masks for use in performing the step of registration.

7. The method as set forth in claim 1 further comprising, filling in image regions associated with the foreground regions of each of the combination masks based upon the step of estimating.

8. The method as set forth in claim 1 further comprising, determining intensity values for anti-aliased pixels located in transition regions of combination masks.

9. The method as set forth in claim 1 wherein the step of estimating includes performing a statistical filtering with respect to pixels in the combination masks, and assigning a discrete intensity value to each of the pixels.

10. The method as set forth in claim 1 wherein the intensity is based upon one of (a) grayscale, (b) a plurality of color channels, and (c) n-channel multi-spectral images.

11. The method as set forth in claim 1 further comprising, rejecting the acquired runtime image based upon information from the registration process that indicates the acquired runtime image exceeds predetermined registration parameters.

12. The method as set forth in claim 1 further comprising, based upon the golden template image, performing a golden template comparison (GTC) with respect to the acquired runtime image.

13. The method as set forth in claim 1 further comprising, generating, based upon at least some of the combination masks, weighting masks that are employed for use in association with an inspection task.

14. The method as set forth in claim 1 wherein the step of estimating is performed on some or all of foreground regions for some or all of the combination masks.

15. A system for generating golden template images in a vision system to inspect an acquired runtime image of an object with a multi-layer printed pattern comprising:

a registration tool that performs registration on runtime images using registration models each trained on respective canonical layer masks, and that outputs poses;

a tool constructed and arranged to (a) generate a plurality of warped layer masks based upon the poses and (b) compute a plurality of combination masks based upon differing combinations of the warped layer masks;

an estimation process that estimates intensity of regions for the combination masks; and

a blending process that blends the estimated intensity values associated with the combination masks to generate a golden template image.

16. The system as set forth in claim 15 wherein the estimation process estimates the intensity of an overall background region.

17. The system as set forth in claim 15 wherein the warped layer masks define a fuzzy edge that, in turn, provide combination masks with fuzzy edges in the step of computing.

18. The system as set forth in claim 15 wherein the warped layer masks are warped based upon distortion within a training image.

19. The system as set forth in claim 15 wherein the warped layer masks are warped based upon distortion in the runtime image currently acquired.

20. The system as set forth in claim 15 further comprising pre-warped distortion information to warp the canonical layer masks for use in performing the registration.

21. The system as set forth in claim 15 wherein image regions associated with the foreground regions of each of the combination masks are filled-in for the estimation process.

22. The system as set forth in claim 15 wherein the estimation process determines intensity values for anti-aliased pixels located in transition regions of combination masks.

23. The system as set forth in claim 15 the intensity values for pixels in combination masks are determined with a statistical filtering, and a discrete intensity value is assigned to each of the pixels.

24. The system as set forth in claim 15 wherein the intensity is based upon one of (a) grayscale, (b) a plurality of color channels, and n-channel multi-spectral images.

25. The system as set forth in claim 15 wherein the registration tool comprises a tool that performs an affine transform between the canonical layer masks and the acquired runtime image.

26. The system as set forth in claim 15 wherein the multi-layer printed pattern of the acquired image comprises at least one of a sprayed, sputtered, rolled, laser-printed, inkjet printed, lithographed, photo-resistive exposed, dry transferred and dithered image.