Machine vision method and apparatus for determining the position of generally rectangular devices using boundary extracting features

• US 5,933,523 A
• Filed: 03/18/1997
• Issued: 08/03/1999
• Est. Priority Date: 03/18/1997
• Status: Expired due to Term

• Current Assignee: Cognex Corporation
• Original Assignee: Cognex Corporation
• Inventors: Motoyama, Toshimitsu, Drisko, Robert, Bachelder, Ivan A.
• Primary Examiner(s): Boudreau, Leo H.
• Assistant Examiner(s): MEHTA, BHAVESH M
• Application Number:
  US08/819,875
• Time in Patent Office: 868 Days
• Field of Search: 364/570, 382/286, 382/289, 382/292, 382/141, 382/145, 382/146-147, 382/151, 382/199, 382/276-277, 382/293, 382/291, 382/295-296
• US Class Current: 382/151
• CPC Class Codes: G06T 2207/30141 Printed circuit board [PCB]
• CPC Class Codes: G06T 7/70 Determining position or orientation of objects or cameras camera calibration G06T7/80
• CPC Class Codes: G06V 10/243 by compensating for image skew or non-uniform image deformations

Abstract

Determining the position of a generally rectangular device is achieved by performing subsampling of image data, thresholding of the subsampled data, extraction of boundary features based on the output of the thresholding, calculation of correspondence between extracted boundary features and the generally rectangular device, and a final fit against a model of the generally rectangular device. Correspondence between the boundary features of the image and the generally rectangular device is performed by locating points in the extracted boundary feature which may correspond to corners of the generally rectangular device, removing those points which correspond to known objects other than the generally rectangular device, removing those points likely to belong to unknown objects, and fitting the points to an input model of the generally rectangular device to determine its position.

Claims

1. A method for identifying and locating the boundaries of a generally rectangular device, said method comprising:

detecting, from a two-dimensional image including a pixel representation of said device, a boundary of said device by locating multiple boundary points lying on a boundary of an object within said two-dimensional image and by representing each of said multiple boundary points with a respective set of first and second cartesian boundary coordinate values respectively defining positions along first and second intersecting axes lying in a plane in common with said two-dimensional image, each of said first and second intersecting axes having an angular position substantially 45°

displaced from a straight line estimated to be parallel to an edge of said device; and

identifying corner points of said device by determining which ones of said respective sets of first and second cartesian boundary coordinate values include at least one of a highest and a lowest of said first cartesian boundary coordinate values and a highest and a lowest of said second cartesian boundary coordinate values.

2. A method according to claim 1, further comprising determining which of said identified corner points is incorrectly identified as a corner point based upon geometric features calculated based upon said identified corner points and compared against model constraints.

3. The method according to claim 2, further comprising determining which of said identified corner points is incorrectly identified as a corner point based upon known information concerning an object other than said device.

4. The method according to claim 1, wherein said two-dimensional image comprises a thresholded image.

5. The method according to claim 4, wherein said two-dimensional thresholded image comprises a front-lit representation of a rectangular surface mount device having chip lead portions at an end thereof, said surface mount device being held by a nozzle.

6. The method according to claim 4, wherein said multiple boundary points intersect a boundaries of plural regions of said device.

7. The method according to claim 6, wherein said detecting of said boundary of said device comprises running a boundary tracker tool.

8. The method according to claim 6, wherein said detecting of said boundary of said device comprises running a region image processing tool.

9. The method according to claim 1, further comprising fitting the corner points against an input model representation of said device to determine coarse part location information concerning said device.

10. The method according to claim 9, wherein said fitting comprises using a three-parameter least squares fit algorithm.

11. The method according to claim 9, wherein said fitting comprises using a rigid-body least squares fit algorithm performing a two-dimensional translation and a rotation.

12. The method according to claim 1, wherein:

said corner points are identified using the following algorithm;

if x_min >

x_Ei

x_min >

x_Ei

minX_-- i=i

else if x_max <

x_Ei

x_max =x_Ei

maxX_-- i=i

if y_min >

y_Ei

y_min >

y_Ei

minY_-- i=i

else if y_max <

y_Ei

y_max =y_Ei

maxY_-- i=i;

wherein x_Ei =x_i -y_i and y_Ei =x_i +y_i, where x_i is an x axis coordinate of a given boundary point i, and y_i is a y axis coordinate of the same boundary point.

13. The method according to claim 1, further comprising ascertaining said first and second intersecting axes, said ascertaining comprising determining an orientation and location of the generally rectangular device.

14. The method according to claim 1, further comprising ascertaining said first and second intersecting axes, said ascertaining comprising determining a pose of the generally rectangular device.

15. Computer-readable media tangibly embodying a program executable by a machine to identify and locate the boundaries of a generally rectangular device, said program performing functions comprising:

detecting, from a two-dimensional image including a pixel representation of said device, a boundary of said device by locating multiple boundary points lying on a boundary of an object within said two-dimensional image and by representing each of said multiple boundary points with a respective set of first and second cartesian boundary coordinate values respectively defining positions along first and second intersecting axes lying in a plane in common with said two-dimensional image, each of said first and second intersecting axes having an angular position substantially 45°

displaced from a straight line estimated to be parallel to an edge of said device; and

identifying corner points of said device by determining which ones of said respective sets of first and second cartesian boundary coordinate values include at least one of a highest and a lowest of said first cartesian boundary coordinate values and a highest and a lowest of said second cartesian boundary coordinate values.

16. The media according to claim 15, said functions further comprising determining which of said identified corner points is incorrectly identified as a corner point based upon geometric features calculated based upon said identified corner points and compared against model constraints.

17. The media according to claim 16, said functions further comprising determining which of said identified corner points is incorrectly identified as a corner point based upon known information concerning an object other than said device.

18. The media according to claim 15, wherein said two-dimensional image comprises a thresholded image.

19. The media according to claim 18, wherein said two-dimensional thresholded image comprises a front-lit representation of a rectangular surface mount device having chip lead portions at an end thereof, said surface mount device being held by a nozzle.

20. The media according to claim 18, wherein said multiple boundary points intersect boundaries of plural regions of said device.

21. The media according to claim 20, wherein said detecting of said boundary of said device comprises running a boundary tracker tool.

22. The media according to claim 20, wherein said detecting of said boundary of said device comprises running a region image processing tool.

23. The media according to claim 15, said functions further comprising fitting the corner points against an input model representation of said device to determine coarse part location information concerning said device.

24. The media according to claim 23, wherein said fitting comprises using a three-parameter least squares fit algorithm.

25. The media according to claim 23, wherein said fitting comprises using a rigid-body least squares fit algorithm performing a two-dimensional translation and a rotation.

26. The media according to claim 15, wherein:

said corner points are identified using the following algorithm;

if x_min >

x_Ei

x_min >

x_Ei

minX_-- i=i

else if x_max <

x_Ei

x_max =x_Ei

maxX_-- i=i

if y_min >

y_Ei

y_min =y_Ei

minY_-- i=i

else if y_max <

y_Ei

y_max =y_Ei

maxY_-- i=i;

wherein x_Ei =x_i -y_i and y_Ei =x_i +y_i, where x_i is an x axes coordinate of a given boundary point i, and y_i is a y axis coordinate of the same boundary point.

27. The media according to claim 15, said functions further comprising ascertaining said first and second intersecting axes, said ascertaining comprising determining an orientation and location of the generally rectangular device.

28. The media according to claim 15, said functions comprising ascertaining said first and second intersecting axes, said ascertaining comprising determining a pose of the generally rectangular device.

29. A system for identifying and locating the boundaries of a generally rectangular device, said system comprising:

means for detecting, from a two-dimensional image including a pixel representation of said device, a boundary of said device by locating multiple boundary points lying on a boundary of an object within said two-dimensional image and by representing each of said multiple boundary points with a respective set of first and second cartesian boundary coordinate values respectively defining positions along first and second intersecting axes lying in a plane in common with said two-dimensional image, each of said first and second intersecting axes having an angular position substantially 45°

displaced from a straight line estimated to be parallel to an edge of said device; and

means identifying corner points of said device by determining which ones of said respective sets of first and second cartesian boundary coordinate values include at least one of a highest and a lowest of said first cartesian boundary coordinate values and a highest and a lowest of said second cartesian boundary coordinate values.

30. The system according to claim 29, further comprising means for determining which of said identified corner points is incorrectly identified as a corner point based upon geometric features calculated based upon said identified corner points and compared against model constraints.

31. The system according to claim 30, further comprising means for determining which of said identified corner points is incorrectly identified as a corner point based upon known information concerning an object other than said device.

32. The system according to claim 29, wherein said two-dimensional image comprises a thresholded image.

33. The system according to claim 32, wherein said two-dimensional thresholded image comprises a front-lit representation of a rectangular surface mount device having chip lead portions at an end thereof, said surface mount device being held by a nozzle.

34. The system according to claim 32, wherein said multiple boundary points intersect boundaries of plural regions of said device.

35. The system according to claim 34, wherein said detecting of said boundary of said device comprises running a boundary tracker tool.

36. The system according to claim 34, wherein said detecting of said boundary of said device comprises running a region image processing tool.

37. The system according to claim 29, further comprising fitting the corner points against an input model representation of said device to determine coarse part location information concerning said device.

38. The system according to claim 37, wherein said fitting comprises using a threeparameter least squares fit algorithm.

39. The system according to claim 37, wherein said fitting comprises using a rigidbody least squares fit algorithm performing a two-dimensional translation and a rotation.

40. The system according to claim 29, wherein:

said corner points are identified using the following algorithm;

if x_min >

x_Ei

x_min >

x_Ei

minX_-- i=i

else if x_max <

x_Ei

x_max =x_Ei

maxX_-- l i=i

if y_min >

y_Ei

y_min =y_Ei

minY_-- i=i

else if y_max <

_Ei

y_max =y_Ei

maxY_-- i=i;

wherein x_Ei =x_i -y_i and y_Ei =x_i +y_i, where x_i is an x axis coordinate of a given boundary point i, and y_i is a y axis coordinate of the same boundary point.

41. The system according to claim 29, further comprising ascertaining said first and second intersecting axes, said ascertaining comprising determining an orientation and location of the generally rectangular device.

42. The system according to claim 29, further comprising ascertaining said first and second intersecting axes, said ascertaining comprising determining a pose of the generally rectangular device.