Method and apparatus for calibrating an image acquisition system

• US 6,798,925 B1
• Filed: 12/22/2000
• Issued: 09/28/2004
• Est. Priority Date: 12/22/2000
• Status: Active Grant

• Current Assignee: Cognex Corporation
• Original Assignee: Cognex Corporation
• Inventors: Wagman, Adam
• Primary Examiner(s): Chang, Jon
• Assistant Examiner(s): TUCKER, WESLEY J
• Application Number:
  US09/747,874
• Time in Patent Office: 1,376 Days
• Field of Search: 382/296, 382/141, 382/159, 382/181, 382/287, 382/142, 382/143, 382/145, 382/147, 382/148, 382/149, 382/151, 382/152, 356/614-616, 356/620, 356/625
• US Class Current: 382/287
• CPC Class Codes: G06T 7/80 Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration

Abstract

A single fiducial mark with at least one characteristic of known value is placed on one or more objects that are examined by a machine vision system. Alternatively, a pre-existing object feature that has a known characteristic can be used. An image acquisition system can be calibrated by examining the objects with a vision tool that can align objects with translation and at least one additional degree of freedom and that can measure the value of the characteristic. In general, information measured by the vision tool can be used to infer similar information concerning the calibration. For example, in one embodiment, the single fiducial mark has a precise dimension and the vision tool can measure the dimension in order to calibrate the scale of the system. In another embodiment, each object to be examined is held in a predetermined position by a fixture on the materials handling system and a fiducial mark is placed on each object. The image acquisition may be calibrated by examining each object, examining objects periodically, or examining objects in another pattern, such as randomly or examining selected objects.

Claims

1. A method for calibrating an image acquisition system with respect to a fixture frame of reference in a machine vision system that has a camera with an image frame of reference, the camera being for obtaining an image of an environment containing an object, the machine vision system having a vision tool capable of aligning location and at least one other non-translational degree of freedom, the method comprising:

(a) placing a single non-rotationally symmetric fiducial mark having at least one precise dimension on the object at a predetermined location on the object and at a predetermined angular orientation with respect to the object;

(b) maintaining a fixed translational and angular relationship between the object and the fixture frame of reference, the camera being stationary with respect to the fixture frame of reference;

(c) using the vision tool to find the fiducial mark in the image so as to determine location and angular orientation of the fiducial mark with respect to the image frame of reference; and

(d) using the fixed relationship indicated in step (b) and the location and angular orientation of the fiducial mark determined in step (c) to calibrate the image acquisition system so as to determine a relationship between measurements and coordinates in the image frame of reference with measurements and coordinates in the fixture frame of reference.

2. The method of claim 1 wherein the vision tool is a multiple pattern normalized correlation tool having a pattern for each object size, and each orientation of each object size.

3. The method of claim 1 wherein the vision tool is a geometrical pattern matching tool.

4. The method of claim 1 wherein step (c) comprises selecting an area of the object having a non-rotationally symmetric feature, and using the vision tool to find the selected area in the image instead of locating a fiducial mark.

5. The method of claim 1 wherein step (c) comprises determining the location and angular orientation of the fiducial mark directly from the outputs of the vision tool.

6. The method of claim 1 wherein step (c) comprises calculating the location and angular orientation of the fiducial mark from the outputs of the vision tool.

7. The method of claim 1 wherein step (b) comprises using a mechanical fixture to maintain a fixed translational and angular relationship between the object and the fixture frame of reference, the camera being stationary with respect to the fixture frame of reference.

8. The method of claim 1 further comprising:

(e) placing a single fiducial mark having at lest one precise dimension on each object of a plurality of objects at respective predetermined locations on the objects.

9. The method of claim 8 further comprising:

(f) maintaining a fixed translational and angular relationship between only some of the plurality of objects and the fixture frame of reference.

10. The method of claim 9 further comprising:

(g) using the fiducial marks on the only some of the plurality of objects to re-calibrate the image acquisition system.

11. The method of claim 8 wherein fiducial mark is non-rotationally symmetric.

12. An apparatus for calibrating an image acquisition system with respect to a fixture frame of reference in a machine vision system that has a camera with an image frame of reference, the camera being for obtaining an image of an environment containing an object, the machine vision system having a vision tool capable of aligning location and at least one other non-translational degree of freedom, the apparatus comprising:

single non-rotationally symmetric fiducial mark having at least one precise dimension placed on the object at a predetermined location on the object and at a predetermined angular orientation with respect to the object;

a fixture for maintaining a fixed translational and angular relationship between the object and the fixture frame of reference, the camera being stationary with respect to the fixture frame of reference;

an image processing system that uses the vision tool to find the fiducial mark in the image so as to determine location and angular orientation with respect to the image frame of reference; and

a calibration system that uses the fixed relationship maintained by the fixture and the location and angular orientation of the fiducial mark determined by the image processing system to calibrate the image acquisition system so as to determine a relationship between measurements and coordinates in the image frame of reference with measurements and coordinates in the fixture frame of reference.

13. The apparatus of claim 12 wherein the vision tool is a multiple pattern normalized correlation tool having a pattern for each object size, and each orientation of each object size.

14. The apparatus of claim 12 wherein the vision tool is a geometrical pattern matching tool.

15. The apparatus of claim 12 wherein the image processing system comprises a mechanism for selecting an area of the object and wherein the vision tool finds the selected area in the image instead of finding a fiducial mark.

16. The apparatus of claim 12 wherein the image processing system comprises a mechanism that determines the location and angular orientation of the fiducial mark directly from the outputs of the vision tool.

17. The apparatus of claim 12 wherein the image processing system comprises a mechanism that calculates the location and angular orientation of the fiducial mark from the outputs of the vision tool.

18. The apparatus of claim 12 wherein the fixture comprises mechanical means to maintain a fixed translational and angular relationship between the object and the fixture frame of reference, the camera being stationary with respect to the fixture frame of reference.

19. The apparatus of claim 12 further comprising a single fiducial mark having at least one precise dimension located on each of a plurality of objects at respective predetermined locations on the objects.

20. The apparatus of claim 19 further comprising a plurality of fixtures for maintaining a fixed translational and angular relationship between at least some of the plurality of objects and the fixture frame of reference.

21. The apparatus of claim 20 further comprising a recalibration system that uses the fiducial marks on the at least some of the plurality of objects to re-calibrate the image acquisition system.

22. The apparatus of claim 19 wherein each fiducial mark is non-rotationally symmetric.

23. A method for calibrating an image acquisition system with respect to a fixture frame of reference in a machine vision system that has a camera with an image frame of reference, the camera being for obtaining an image of an environment containing an object, the camera being in stationary relationship with the object the machine vision system having a vision tool capable of aligning location and at least on other non-translational degree of freedom, the method comprising:

(a) electing a feature having at least one characteristic of known value on the object;

(b) using the vision tool to find the feature in the image so as to determine the value of the at least one characteristic with respect to the image frame of reference; and

(c) using the value of the at least one characteristic to calibrate the image acquisition system so as to determine at least a relationship between pixel size in the image frame of reference with measurements in the fixture frame of reference.

24. The method of claim 23 wherein the feature is a fiducial mark placed on the object in order to calibrate the image acquisition system.

25. The method of claim 24 wherein the fiducial mark is non-rotationally symmetric.

26. The method of claim 23 wherein the feature is a pre-existing feature that naturally occurs on the object.

27. The method of claim 23 wherein the characteristic of known value is a precise dimension and the method calibrates a scale of the image acquisition system.

28. The method of claim 23 wherein the feature is not rotationally symmetric and the characteristic of known value is an angular orientation and the method calibrates an angular orientation of the image acquisition system.

29. The method of claim 23 wherein the feature is not rotationally symmetric and has at least two characteristics of known value including a precise dimension and an angular orientation and the method calibrates both a scale and an angular orientation of the image acquisition system.

30. The method of claim 23 wherein step (b) comprises using a mechanism for selecting an area of the object and then using the vision tool to find the selected area in the image instead of locating a fiducial mark.

31. An apparatus for calibrating an image acquisition system with respect to a fixture frame of reference in a machine vision system that has a camera with an image frame of reference, the camera being for obtaining an image of an environment containing as object, the camera being in stationary relationship with the object, the machine vision system having a vision tool capable of aligning location and at least one other non-translational degree of freedom, the apparatus comprising:

a mechanism that selects a feature having at least one characteristic of known value on the object;

a vision tool that finds the feature in the image so as to determine the value of the at least one characteristic with respect to the image frame of reference; and

an object processing system that uses the value of the at least one characteristic to calibrate the image acquisition system so as to determine at least a relationship between pixel size in the image frame of reference with measurements in the fixture frame of reference.

32. The apparatus of claim 31 wherein the feature is a fiducial mark place on the object in order to calibrate the image acquisition system.

33. The apparatus of claim 32 wherein the fiducial mark is non-rotationally symmetric.

34. The apparatus of claim 31 wherein the feature is a pre-existing feature that naturally occurs on the object.

35. The apparatus of claim 31 wherein the characteristic of known value is a precise dimension ad the method calibrates a scale of the image acquisition system.

36. The apparatus of claim 31 wherein the feature is not rotational symmetric and the characteristic of known value is an angular orientation and the method calibrates an angular orientation of the image acquisition system.

37. The apparatus of claim 31 wherein the feature is not rotationally symmetric and has at least two characteristics of known value including a precise dimension and an angular orientation and the method calibrates both a scale and an angular orientation of the image acquisition system.

38. The apparatus of claim 31 wherein the vision tool is cooperative with a mechanism that selects an area of the object and wherein the vision tool searches for the selected area.

39. A machine vision system for calibrating an image acquisition system with respect to a fixture frame of reference, the machine vision system comprising:

a camera for obtaining an image of an environment containing an object, the object being in stationary relationship with the camera while the camera obtains an image of the object;

a vision tool capable of aligning location and at least one other non-translational degree of freedom;

a single non-rotationally symmetric fiducial mark having at least one precise dimension placed on the object at a predetermined location on the object and at a predetermined angular orientation with respect to the object;

a fixture for maintaining a fixed translational and angular relationship between the object and the fixture frame of reference, the camera being stationary with respect to the fixture frame of reference;

a search system that uses the vision tool to find the fiducial mark in the image so as to determine its location and angular orientation with respect to the image frame of reference; and

a calibration system that uses the fixed relationship maintained by the fixture, and the location and angular orientation of the fiducial mark determined by the search system, to calibrate the image acquisition system so as to determine a relationship between measurements and coordinates in the image frame of reference with measurements and coordinates in the fixture frame of reference.