Bandwidth reduction of multichannel images for machine vision

• US 6,141,033 A
• Filed: 05/15/1997
• Issued: 10/31/2000
• Est. Priority Date: 05/15/1997
• Status: Expired due to Term

• Current Assignee: Cognex Corporation
• Original Assignee: Cognex Corporation
• Inventors: Romanik, Philip B., Michael, David J.
• Primary Examiner(s): Kelley, Chris S.
• Assistant Examiner(s): PHILIPPE, GIMS S
• Application Number:
  US08/856,920
• Time in Patent Office: 1,265 Days
• Field of Search: 348/25, 348/87, 348/384, 348/390, 382/141, 382/144, 382/145, 382/146, 382/147, 382/148, 382/149, 382/232, 382/235, 382/169, 382/209, 382/151, 382/302
• US Class Current: 348/25
• CPC Class Codes: G06V 10/28 Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns
• CPC Class Codes: H04N 19/30 using hierarchical techniques, e.g. scalability H04N19/63 takes precedence
• CPC Class Codes: H04N 19/85 using pre-processing or post-processing specially adapted for video compression

Abstract

The invention provides methods and apparatus for reducing the bandwidth of a multichannel image. In one aspect, the methods and apparatus call for acquiring a multichannel training image representing a training scene. Weighting factors for the respective channels are determined based on the contrast at corresponding locations in that multichannel training image. A reduced bandwidth runtime image is generated from the multichannel runtime images as a function of (i) the weighting factors determined from the training image and (ii) a multichannel image representing the runtime scene.

Claims

1. A method of reducing bandwidth of a multichannel image that has multiple channels each representing a respective spectral component of a scene, the method comprising the steps of:

A. identifying a calibration location in a multichannel training image;

B. determining a contrast within each channel of the multichannel training image at a location corresponding to the calibration location;

C. determining a weighting factor for each channel of the multichannel training image as a function of the contrast within the channel at the location corresponding to the calibration location; and

D. generating, from a multichannel runtime image, a first reduced bandwidth runtime image but that has fewer channels than the multichannel runtime image, the reduced bandwidth runtime image being generated as a first function of (i) the multichannel runtime image, and (ii) the weighting factors for the channels of the multichannel training image.

2. A method according to claim 1, wherein step (C) includes the step of generating the weighting factors so as to emphasize a contrast at the calibration location in the training image.

3. A method according to claim 2, wherein the first function utilized in step (D) is a linear function of (i) the channels of the multichannel runtime image, and (ii) the weighting factors for each channel of the multichannel training image.

4. A method of reducing bandwidth of a multichannel image that has multiple channels each representing a respective spectral component of a scene, the method comprising the steps of:

A. identifying a calibration location in a multichannel training image;

B. determining a contrast within each channel of the multichannel training image at a location corresponding to the calibration location;

C. determining a weighting factor for each channel of the multichannel training image as a function of the contrast within the channel at the location corresponding to the calibration location;

said weighting factors being generated so as to emphasize a contrast at the calibration location in the training image; and

D. generating, from a multichannel runtime image, a first reduced bandwidth runtime image but that has fewer channels than the multichannel runtime image, the reduced bandwidth runtime image being generated as a first function of (i) the multichannel runtime image, and (ii) the weighting factors for the channels of the multichannel training image,

wherein the first function utilized in step (D) is represented by the following relationship;

space="preserve" listing-type="equation">I.sub.1 =k.sub.1 C.sub.1 +K.sub.2 C.sub.2 + . . . K.sub.n C.sub.n

where

I₁ is the first reduced bandwidth runtime image;

C₁ -Cn are channels of the multichannel runtime image; and

k₁ -kn are weighting factors for the channel C₁ -C_n.

5. A method according to claim 4, wherein step (C) includes the step of determining the weighting factors so that, together, their sum is a selected value.

6. A method according to claim 5, wherein step (C) includes the step of determining the weighting factors so that, together, their sum is substantially one.

7. A method for reducing bandwidth of a multichannel image that has multiple channels each representing a respective spectral component of a scene, the method comprising the steps of:

A. identifying a calibration location in a multichannel training image;

B. determining a contrast within each channel of the multichannel training image at a location corresponding to the calibration location;

C. determining a weighting factor for each channel of the multichannel training image as a function of the contrast within the channel at the location corresponding to the calibration location; and

D. generating, from a multichannel runtime image, a first reduced bandwidth runtime image but that has fewer channels than the multichannel runtime image, the reduced bandwidth runtime image being generated as a first function of (i) the multichannel runtime image, and (ii) the weighting factors for the channels of the multichannel training image, said first function being a linear function of (i) the channels of the multichannel runtime image, and (ii) the weighting factors for each channel of the multichannel training image,

wherein step (C) further includes the steps of;

generating, from the multichannel training image, a first reduced bandwidth training image, the first reduced bandwidth training image being generated using the first function;

generating, from the multichannel training image, a second reduced bandwidth training image, the second reduced training image being generated using a second function that differs from the first function;

comparing a contrast at the calibration location in the first reduced bandwidth training image with that at the calibration location in the second reduced bandwidth training image; and

utilizing, in step (D) a selected one of the first and second functions, based on a result of the comparing step.

8. A method according to claim 7, wherein the second function is a hue function.

9. A method according to claim 8, comprising the step of determining a hue saturation in the multichannel training image and, depending on a degree of hue saturation, generating the second reduced bandwidth image as function of a hue of the corresponding multichannel image.

10. A method according to claim 9, wherein the second function is represented by the relationship:

##EQU8##

space="preserve" listing-type="equation">I.sub.2 =a tan 2(x,y)

where

I₂ is the second reduced bandwidth image;

C₁ -C_n are channels of the multichannel image from which the first reduced bandwidth image is generated; and

atan2 is an arctangent function.

11. A method of bandwidth reduction for machine vision image acquisition, the method comprising the steps of:

A. identifying a calibration location in a multichannel training image that has multiple channels, each representing a respective spectral component of a training scene;

B. determining a contrast within each channel of the multichannel training image at a location corresponding to the calibration location;

C. determining a weighting factor for each channel of the multichannel training image as a function of the contrast within that channel at the location corresponding to the calibration location; and

D. adjusting an optical element to alter a spectral content of a runtime scene and inputting an image of that scene with a second image acquisition device that represents the image with fewer channels than the multichannel training image.

12. A method according to claim 11, wherein the optical element is an optical filter that alters a spectral content of the image acquired by the second image acquisition device.

13. A method according to claim 12, wherein the optical element is an illumination source that alters a spectral content of the image acquired by the second image acquisition device.

14. A method according to any of claims 12 and 13 wherein the step (D) includes the step of imaging the runtime scene with a second image acquisition device that acquires a gray-scale image.

15. A method according to any of claims 12 and 13, wherein step (D) includes the step of imaging the runtime scene with a second image acquisition device that acquires a gray-scale image.

16. A machine vision method of analyzing, with a gray-scale machine vision tool, a multichannel image that has multiple channels, each representing a respective spectral component of a scene, the method comprising the steps of:

A. determining, from a multichannel training image representing a training scene, a mapping function;

B. generating, from a multichannel runtime image representing a runtime scene, a reduced bandwidth runtime image that represents the runtime scene but that has fewer channels than the multichannel runtime image, the reduced bandwidth runtime image being generated as a function of (i) the multichannel runtime image, and (ii) the mapping function; and

C. applying the reduced bandwidth runtime image to the gray-scale machine vision tool.

17. A method according to claim 16, wherein step (A) includes the steps of

identifying a calibration location in a multichannel training image representing a training scene;

determining a contrast within each channel of the multichannel training image at a location corresponding to the calibration location; and

determining a weighting factor for each channel of the multichannel training image as a function of the contrast within that channel at the location corresponding to the calibration location.

18. A method according to claim 17, wherein step (B) includes the step of generating the reduced bandwidth runtime image as a function of (i) the multichannel runtime image, and (ii) the weighting factors for the channels of the multichannel training image.

19. A method according to claim 16, wherein step (A) includes the steps of

identifying one or more calibration locations in a multichannel training image representing a training scene;

generating a look-up table as a function of the image values in one or more of the channels of the multichannel training image at each of the calibration locations.

20. A method according to claim 19, wherein step (B) includes the step of generating at least selected image values in the reduced bandwidth runtime image as a function of (i) corresponding image values in the multichannel runtime image, and (ii) corresponding values in the look-up table.

21. An article of manufacture comprising a computer usable medium embodying program code for causing a digital data processor to carry out a method of reducing bandwidth of a multichannel image that has multiple channels each representing a respective spectral component of a scene, the method comprising the steps of:

A. identifying a calibration location in a multichannel training image representing a training scene;

B. determining a contrast within each channel of the multichannel training image at a location corresponding to the calibration location;

C. determining a weighting factor for each channel of the multichannel training image as a function of the contrast within the channel at the location corresponding to the calibration location; and

D. generating, from a multichannel runtime image representing a runtime scene, a first reduced bandwidth runtime image but that has fewer channels than the multichannel runtime image, the reduced bandwidth runtime image being generated as a first function of (i) the multichannel runtime image, and (ii) the weighting factors for the channels of the multichannel training image.

22. An article of manufacture according to claim 21, wherein step (C) of the method includes the step of generating the weighting factors so as to emphasize a contrast at the calibration location in the training image.

23. An article of manufacture comprising a computer usable medium embodying program code for causing a digital data processor to carry out a method of bandwidth reduction for machine vision image acquisition, the method comprising the steps of:

A. inputting with a first image acquisition device a multichannel training image representing a training scene and having multiple channels that each represent a respective spectral component of the training scene;

B. identifying a calibration location in the multichannel training image;

C. determining a contrast within each channel of the multichannel training image at a location corresponding to the calibration location;

D. determining a weighting factor for each channel of the multichannel training image as a function of the contrast within that channel at the location corresponding to the calibration location;

E. adjusting an optical element to alter a spectral content of a runtime scene and inputting an image with a second image acquisition device that represents that scene with a reduced bandwidth signal having fewer channels than the multichannel training image.

24. An article of manufacture according to claim 23, wherein the optical element is an optical filter that alters a spectral content of the image acquired by the second image acquisition device.

25. An article of manufacture comprising a computer usable medium embodying program code for causing a digital data processor to carry out a method of analyzing with a gray-scale machine vision tool a multichannel image that represents a scene and that has multiple channels each representing a respective spectral component of the scene, the method comprising the steps of:

A. determining, from a multichannel training image representing a training scene, a mapping function;

B. generating, from a multichannel runtime image representing a runtime scene, a reduced bandwidth runtime image but that has fewer channels than the multichannel runtime image, the reduced bandwidth runtime image being generated as a function of (i) the multichannel runtime image, and (ii) the mapping function; and

C. applying the reduced bandwidth runtime image to the gray-scale machine vision tool.

26. An article of manufacture according to claim 25, wherein step (A) of the method includes the steps of

identifying a calibration location in a multichannel training image representing a training scene;

determining a contrast within each channel of the multichannel training image at a location corresponding to the calibration location; and

determining a weighting factor for each channel of the multichannel training image as a function of the contrast within the channel at the location corresponding to the calibration location.

27. An article of manufacture according to claim 25, wherein step (A) of the method includes the steps of

identifying one or more calibration locations in a multichannel training image representing a training scene;

generating a look-up table as a function of the image values in one or more of the channels of the multichannel training image at each of the calibration locations.

28. An apparatus for reducing bandwidth of a multichannel image that represents a scene and that has multiple channels each representing a respective spectral component of the scene, the apparatus comprising:

A. means for identifying a calibration location in a multichannel training image representing a training scene;

B. means for determining a contrast within each channel of the multichannel training image at a location corresponding to the calibration location;

C. means for determining a weighting factor for each channel of the multichannel training image as a function of the contrast within the channel at the location corresponding to the calibration location; and

D. means for generating, from a multichannel runtime image representing a runtime scene, a first reduced bandwidth runtime image runtime scene but that has fewer channels than the multichannel runtime image, the reduced bandwidth runtime image being generated as a first function of (i) the multichannel runtime image, and (ii) the weighting factors for the channels of the multichannel training image.

29. An apparatus according to claim 28, wherein the means for determining generates the weighting factors so as to emphasize a contrast at the calibration location in the training image.

30. An apparatus for reducing the bandwidth of a runtime image based on analysis of a training image, the apparatus comprising:

A. means for identifying a calibration location in a multichannel training image having multiple channels that each represent a respective spectral component of a training scene;

B. means for determining a contrast within each channel of the multichannel training image at a location corresponding to the calibration location;

C. means for determining a weighting factor for each channel of the multichannel training image as a function of the contrast within that channel at the location corresponding to the calibration location;

D. means for adjusting an optical element to alter a spectral content of a runtime scene and inputting that image with a second image acquisition device that represents that scene with a reduced bandwidth image having fewer channels than the multichannel training image.

31. An apparatus according to claim 30, wherein the optical element is an optical filter that alters a spectral content of the image acquired by the second image acquisition device.

32. A machine vision apparatus for analyzing with a gray-scale machine vision tool a multichannel signal that represents a scene and that has multiple channels each representing a respective spectral component of the scene, the method comprising the steps of:

A. means for determining, from a multichannel training image representing a training scene, a mapping function;

B. means for generating, from a multichannel runtime image representing a runtime scene, a reduced bandwidth runtime that represents the runtime scene but that has fewer channels than the multichannel runtime image, the reduced bandwidth runtime image being generated as a function of (i) the multichannel runtime image, and (ii) the mapping function; and

C. means for applying the reduced bandwidth runtime image to the gray-scale machine vision tool.

33. An apparatus according to claim 32, wherein the means for determining includes

means for identifying a calibration location in a multichannel runtime image representing a training scene;

means for determining a contrast within each channel of the multichannel training image at a location corresponding to the calibration location; and

means for determining a weighting factor for each channel of the multichannel training image as a function of the contrast within the channel at the location corresponding to the calibration location.

34. An apparatus according to claim 32, wherein the means for determining includes

means for identifying one or more calibration locations in a multichannel training image representing a training scene;

means for generating a look-up table as a function of the image values in one or more of the channels of the multichannel training image at each of the calibration locations.