System and method for detecting flaws in objects using machine vision

• US 8,103,085 B1
• Filed: 09/25/2007
• Issued: 01/24/2012
• Est. Priority Date: 09/25/2007
• Status: Active Grant

• Current Assignee: Cognex Corporation
• Original Assignee: Cognex Corporation
• Inventors: Zadeh, Ali Moghaddam
• Primary Examiner(s): Tucker, Wesley
• Assistant Examiner(s): WOLDEMARIAM, AKLILU K
• Application Number:
  US11/860,800
• Time in Patent Office: 1,582 Days
• Field of Search: 382/141, 356/237.1, 356/124, 356/237, 356/612, 348/126, 348/127
• US Class Current: 382/141
• CPC Class Codes: G06T 2207/30164 Workpiece; Machine component
• CPC Class Codes: G06T 7/001 using an image reference approach
• CPC Class Codes: G06T 7/33 using feature-based methods

Abstract

This invention overcomes the disadvantages of the prior art by providing a system and method for flexibly detecting flaws in the acquired runtime/live images of objects based upon an inspection process that employs a training or model image of the object. This system and method enables predetermined flaws, and other predetermined features, within predetermined tolerances, to be disregarded as a flaw at inspection time. Typically, flexible flaw detection and repositioning of the model image occurs locally with respect to predetermined flaws and imperfections, and is undertaken after the image is acquired and the model image has undergone a global affine transformation/positioning with respect to the runtime image. Subsequent inspection then occurs with a locally repositioned model image that subsumes any imperfection or variation that is to be disregarded.

Claims

1. A system for flexibly detecting flaws in an object comprising:

an imaging device having a processor that performs a global registration of a model image with respect to a runtime image;

a flexible flaw detection and repositioning procedure within the processor that identifies a displacement of predetermined image features in the model image with respect to the runtime image and that computes a relative displacement therebetween so that local positioning of predetermined image features on the model image can be transformed with respect to the runtime image to determine if the relative displacement results in a better fit for the model image relative to the runtime image and, if not, the relative displacement is marked invalid; and

an inspection procedure that performs inspection of the runtime image using the model image subsequent to the transformation of the predetermined image features.

2. The system as set forth in claim 1 wherein the flexible flaw detection and repositioning procedure is constructed and arranged to connect model image edge points to derive model edge segments and to connect runtime image edge points to derive runtime edge segments and to match the model edge segments with corresponding of the runtime edge segments.

3. The system as set forth in claim 2 wherein the flexible flaw detection and repositioning procedure is constructed and arranged to compute displacement between an enclosed segment feature in the model image and an enclosed segment feature in the runtime image.

4. The system as set forth in claim 3 wherein the flexible flaw detection and repositioning procedure is constructed and arranged to compute respective displacements between pixels on a model image edge and pixels on a runtime image edge.

5. The system as set forth in claim 4 further comprising, a location transformation procedure that, for each model image pixel employs the displacements of nearby model edge pixels to determine a displacement thereof with respect to a corresponding runtime image pixel.

6. The system and set forth in claim 5 wherein the location adjustment transformation procedure is constructed and arranged to divide the model image and the runtime image, respectively, into bins and to average the displacements within each of the bins.

7. The system as set forth in claim 6 wherein the inspection procedure comprises at least one of area-based inspection and edge-segment based inspection.

8. The system as set forth in claim 1 wherein the inspection procedure comprises at least one of area-based inspection and edge-segment based inspection.

9. A method for flexibly detecting flaws in an object comprising the steps of:

implementing one or more processors to perform the following steps;

performing a global registration of a model image with respect to a runtime image;

after global registration, identifying a displacement of predetermined image features in the model image with respect to the runtime image and computing a relative displacement therebetween so that local positioning of predetermined image features on the model image can be transformed with respect to the runtime image and thereafter determining if the relative displacement results in a better fit for the model image relative to the runtime image and, if not, the relative displacement is discarded or marked invalid; and

inspecting the runtime image using the model image subsequent to the transformation of the predetermined image features.

10. The method as set forth in claim 9 wherein the step of computing the relative displacement includes connecting image edge points to derive model edge segments and connecting runtime image edge points to derive runtime edge segments, and matching the model edge segments with corresponding of the runtime edge segments.

11. The method as set forth in claim 10 wherein the step of computing includes determining a displacement between an enclosed segment feature in the model image and an enclosed segment feature in the runtime image.

12. The method as set forth in claim 11 wherein the step of computing includes determining respective displacements between pixels on a model image edge and pixels on a runtime image edge.

13. The method as set forth in claim 12 further comprising, determining, for each model image pixel, the displacements of nearby model edge pixels so as to derive a displacement with respect to a corresponding runtime image pixel.

14. The method and set forth in claim 13 wherein the step of determining the displacement of nearby model edge pixels includes dividing the model image and the runtime image, respectively, into bins and to average the displacements within each of the bins.

15. The system as set forth in claim 14 wherein the step of inspecting comprises at least one of area-based inspection and edge-segment based inspection.

16. The system as set forth in claim 9 wherein the step of inspecting comprises at least one of area-based inspection and edge-segment based inspection.

17. A method for inspecting an object comprising the steps of:

implementing one or more processors to perform the following steps;

performing a global registration of a runtime image of the object with respect to a model image;

performing a flexible local repositioning of at least enclosed segment features and edge features and deriving respective displacement vectors therefor and thereafter determining if the respective displacement vectors result in a better fit for the model image relative to the runtime image and, if not, the respective displacement vectors are discarded or marked invalid;

inspecting the runtime image using the model image with the enclosed segment features and the edge features of the model image being locally repositioned within predetermined threshold values so that flaws falling within the locally repositioned model image are disregarded and flaws falling outside the locally repositioned model image are indicated.