Method for coordinating multiple fields of view in multi-camera

• US 5,768,443 A
• Filed: 12/19/1995
• Issued: 06/16/1998
• Est. Priority Date: 12/19/1995
• Status: Expired due to Term

• Current Assignee: Cognex Corporation
• Original Assignee: Cognex Corporation
• Inventors: Michael, David, Wallack, Aaron
• Primary Examiner(s): COUSO, YON JUNG
• Application Number:
  US08/574,733
• Time in Patent Office: 910 Days
• Field of Search: 382/294, 382/151, 382/152, 382/284, 355/203, 348/159, 348/95
• US Class Current: 382/294
• CPC Class Codes: G01N 21/95 characterised by the material or shape of the object to be examined G01N21/89 - G01N21/91, G01N21/94 take precedence
• CPC Class Codes: G06T 5/80 Geometric correction
• CPC Class Codes: G06T 7/85 Stereo camera calibration

Abstract

A method is provided for use in a multi-camera machine vision system wherein each of a plurality of cameras simultaneously acquires an image of a different portion of an object of interest. The invention makes it possible to precisely coordinate the fields of view of the plurality of cameras so that accurate measurements can be precisely performed across multiple fields of view, even in the presence of image distortion within each field of view. The method includes the steps of, at calibration-time, fixing the plurality of cameras with respect to a substantially rigid dimensionally-stable substrate including a plurality of calibration targets each having a reference feature. For each camera, an image of a calibration target is acquired to provide a plurality of acquired calibration target images. Then a distortion-correction map is generated for each acquired calibration target image. At run-time, for each camera, an image is acquired, at least two of the images including a portion of the object to provide a plurality of partial object images. These partial object images are then transformed by a distortion-correction map to provide a plurality of corrected partial object images. Next, relative displacement information is used to determine the relative displacement of a first point in a first corrected partial object image with respect to a second point in a second corrected partial object image. A combined map can be generated that both corrects image distortion, and transforms local camera coordinates into global coordinates.

Claims

1. A method for precisely coordinating multiple fields of view of a plurality of fixed cameras so as to facilitate determining the distance between features each disposed within a different field of view, the method being especially useful when there is image distortion in each field of view, the method comprising the steps of:

at calibration-time,

fixing a substantially rigid dimensionally-stable substrate including a plurality of calibration targets each having a reference feature, such that a calibration target is within the field of view of each said camera;

acquiring relative displacement information regarding at least the linear displacement of each said reference feature with respect to at least one other said reference feature;

for each camera, acquiring an image of a calibration target to provide a plurality of acquired calibration target images;

generating an image distortion-correction map for each acquired calibration target image, and

at run-time,

for each camera, acquiring an image of a portion of said object to provide a plurality of partial object images;

applying said image distortion-correction map to each partial object image to provide a plurality of corrected partial object images; and

using said relative displacement information to determine the relative displacement of a first point in a first corrected partial object image with respect to a second point in a second corrected partial object image.

2. The method of claim 1, wherein at run-time, the step of acquiring an image includes acquiring an image from each of at least three cameras.

3. The method of claim 1, further including the step of:

at calibration time, acquiring absolute displacement information regarding at least the linear displacement of at least one said reference feature with respect to a reference feature of a mechanical device.

4. The method of claim 3, wherein said substrate is transparent, and the step of acquiring absolute displacement information includes the step of viewing said reference feature of said mechanical device through said substrate.

5. The method of claim 1, wherein the step of acquiring an image of a calibration target includes the step of acquiring an image of an array of dots.

6. The method of claim 1, wherein the step of acquiring an image of a calibration target includes the step of acquiring the image of a grid of lines.

7. The method of claim 1, wherein the step of acquiring an image of a calibration target includes the step of acquiring an image of two non-parallel lines for indicating an origin of said calibration target.

8. The method of claim 1, wherein said step of acquiring relative displacement information includes acquiring relative angular displacement information.

9. The method of claim 1, wherein said step of generating a distortion-correction map includes the step of determining the coefficients of a polynomial.

10. The method of claim 9, wherein said step of determining the coefficients of a polynomial includes the step of determining the coefficients of a fifth-order polynomial.

11. The method of claim 1, wherein said step of generating a distortion correction map includes the step of pre-computing said distortion correction map and storing the results so-computed in a look-up table.

12. The method of claim 1, wherein the step of applying said distortion-correction map includes the step of reading a correction parameter in a look-up table.

13. A method for precisely coordinating multiple fields of view of a plurality of fixed cameras so as to facilitate determining the distance between features each disposed within a different field of view, the method being especially useful when there is image distortion in each field of view, the method comprising the steps of:

at calibration-time,

fixing a plurality of cameras with respect to a calibration plate including a plurality of calibration targets;

for each camera, generating an image distortion-correction map;

for each camera, generating a transformation of camera coordinates into global coordinates; and

at run-time,

for each camera, acquiring an image of a portion of said object to provide a plurality of partial object images;

applying said image distortion-correction map to each partial object image to provide a plurality of corrected partial object images; and

using said transformation of camera coordinates into global coordinates to determine the relative displacement of a first feature point in a first corrected partial object image with respect to a second feature point in a second corrected partial object image.

14. The method of claim 13, further including the step of:

at calibration time, acquiring absolute displacement information regarding at least the linear displacement of at least one said reference feature with respect to a reference feature of a mechanical device.

15. The method of claim 13, wherein said step of generating a distortion-correction map includes the step of determining the coefficients of a polynomial.

16. The method of claim 13, wherein said step of generating a distortion correction map includes the step of pre-computing said distortion correction map and storing the results so-computed in a look-up table.

17. A method for precisely coordinating multiple fields of view of a plurality of fixed cameras so as to facilitate determining the distance between features each disposed within a different field of view, the method being especially useful when there is image distortion in each field of view, the method comprising the steps of:

at calibration-time,

fixing a plurality of cameras with respect to a calibration plate including a plurality of calibration targets;

for each camera, generating an image distortion-correction map using said plurality of calibration targets;

for each camera, generating a transformation of local camera coordinates into global coordinates using said plurality of calibration targets;

combining said image distortion-correction map and said transformation of local camera coordinates into global coordinates to provide a corrected local-to-global map for determining the distance between features each disposed within a different field of view, even when there is image distortion in each field of view; and

at run-time,

for each camera, acquiring an image of a portion of said object to provide a plurality of partial object images;

applying said corrected local-to-global map to determine the relative displacement of a first feature point in a first partial object image with respect to a second feature point in a second partial object image.

18. The method of claim 17, further including the step of:

at calibration time, acquiring absolute displacement information with respect to a reference feature of a mechanical device.

19. The method of claim 17, wherein said step of generating an image distortion-correction map includes the step of determining the coefficients of a polynomial.

20. The method of claim 17, wherein said step of generating an image distortion correction map includes the step of pre-computing said image distortion correction map and storing the results so-computed in a look-up table.