Automatic image adjustment

• US 8,086,029 B1
• Filed: 12/13/2006
• Issued: 12/27/2011
• Est. Priority Date: 12/13/2006
• Status: Active Grant

• Current Assignee: Adobe Inc.
• Original Assignee: Adobe Systems Incorporated (Adobe Inc.)
• Inventors: Chien, Jen-Chan, Hamburg, Mark
• Primary Examiner(s): Tran, Phuoc
• Application Number:
  US11/638,781
• Time in Patent Office: 1,840 Days
• Field of Search: 382/162, 382/167, 382/225, 382/307
• US Class Current: 382/162
• CPC Class Codes: H04N 1/6027 Correction or control of colour gradation or colour contrast H04N1/6058 takes precedence

Abstract

Automatic image adjustment is described, including performing a principal component analysis (PCA) for a plurality of values to determine a principal component, normalizing the principal component to obtain a normalized principal component, and generating an image by mapping the plurality of values using the normalized principal component.

Claims

1. A method, comprising:

performing, by a computer, a principal component analysis (PCA) on an image for a plurality of values of the image to determine a principal component;

normalizing, by the computer, the principal component to obtain a normalized principal component; and

generating, by the computer, another image by mapping the plurality of values using the normalized principal component.

2. The method of claim 1, wherein the normalizing is performed before the generating.

3. The method of claim 1, wherein normalizing comprises performing a linear normalization.

4. The method of claim 3, wherein normalizing comprises solving an equation comprising a sum of a component of the principal component and at least one another component of the principal component set equal to a constant value.

5. The method of claim 1, wherein normalizing comprises performing a non-linear normalization.

6. The method of claim 5, wherein normalizing comprises solving an equation comprising a sum of a first square of a component of the principal component and at least a second square of another component of the principal component set equal to a constant value.

7. The method of claim 5, wherein normalizing comprises solving an equation comprising a sum of a first product of a component of the principal component and a first absolute value of the component and at least a second product of another component of the principal component and a second absolute value of the another component set equal to a constant value.

8. The method of claim 1, wherein mapping comprises generating a grayscale image by mapping the principal component to the plurality of values.

9. The method of claim 1, wherein performing a PCA comprises performing a three-way PCA.

10. The method of claim 1, wherein performing a PCA comprises performing a six-way PCA.

11. The method of claim 1, wherein performing a PCA comprises performing a PCA using a subset of the plurality of values.

12. A method, comprising:

performing a principal component analysis (PCA) on an image to determine a principal component;

applying a plurality of weights based on the principal component to a channel mixer; and

generating a grayscale image from the plurality of weights using the channel mixer.

13. The method of claim 12, wherein performing the PCA comprises performing a three-way PCA.

14. The method of claim 13, further comprising:

determining a red weight using a red function with the PCA, the red function comprising a plurality of red values of the image;

determining a green weight using a green function with the PCA, the green function comprising a plurality of green values of the image;

determining a blue weight using a blue function with the PCA, the blue function comprising a plurality of blue values of the image; and

applying the red weight, the green weight, the blue weight to the channel mixer.

15. The method of claim 14, further comprising:

determining a yellow weight that is a complement of the red weight;

determining a cyan weight that is a complement of the green weight;

determining a magenta weight that is a complement of the blue weight; and

applying the cyan weight, and the magenta weight to the channel mixer.

16. The method of claim 12, further comprising:

normalizing the principal component to obtain a plurality of normalized components of the principal component; and

wherein applying the plurality of weights comprises applying the plurality of weights based on the plurality of normalized components.

17. The method of claim 12, wherein performing the PCA comprises performing a six-way PCA.

18. The method of claim 17, wherein performing the PCA comprises:

determining a red value, a green value, and a blue value for a pixel of the image; and

determining a red function comprising the red value, a green function comprising the green value, and a blue function comprising the blue value.

19. The method of claim 18, further comprising:

determining a yellow function comprising a first minimum of the red value and the green value;

determining a cyan function comprising a first minimum of the green value and the blue value;

determining a magenta function comprising a first minimum of the blue value and the red value;

determining the principal component using the red function, the green function, the blue function, the yellow function, the cyan function, and the magenta function; and

wherein the principal component comprises a red component, a green component, and a blue component, a yellow component, a cyan component, and a magenta component.

20. The method of claim 18, further comprising:

determining a maximum function comprising a maximum value of the red value, the green value, and the blue value;

determining a middle function comprising a middle value of the red value, the green value, and the blue value;

determining a minimum function comprising a minimum value of the red value, the green value, and the blue value;

determining the principal component using the red function, the green function, the blue function, the maximum function, the middle function, and the minimum function; and

wherein the principal component comprises a red component, a green component, and a blue component, a maximum component, a middle component, and a minimum component.

21. The method of claim 20, further comprising:

determining a red weight comprising a first sum of the red component and the maximum component;

determining a yellow weight comprising a second sum of the red component, the green component, the middle component, and the maximum component;

determining a green weight comprising a third sum of the green component and the maximum component;

determining a cyan weight comprising a fourth sum of the green component, the blue component, the middle component and the maximum component;

determining a blue weight comprising a fifth sum of the blue component and the maximum component;

determining a magenta weight comprising a sixth sum of the blue component, the red component, the middle component, and the maximum component; and

applying the red weight, the green weight, the blue weight, the yellow weight, the cyan weight, and the magenta weight to the channel mixer.

22. The method of claim 19, further comprising:

determining a red weight comprising the red component;

determining a yellow weight comprising a first sum of the red component, the green component, and the yellow component;

determining a green weight comprising the green component;

determining a cyan weight comprising a second sum of the green component, the blue component, and the cyan component;

determining a blue weight comprising the blue component;

determining a magenta weight comprising a third sum of the blue component, the red component, and the magenta component; and

applying the red weight, the green weight, the blue weight, the yellow weight, the cyan weight, and the magenta weight to the channel mixer.

23. A system, comprising:

a memory configured to generate a grayscale image; and

a processor configured to;

perform a principal component analysis (PCA) on an image to determine a principal component;

apply a plurality of weights based on the principal component to a channel mixer; and

generate a grayscale image from the plurality of weights using the channel mixer.

24. The system of claim 23, the processor being further configured to:

determine a red weight using a red function with the PCA, the red function comprising a plurality of red values of the image;

determine a green weight using a green function with the PCA, the green function comprising a plurality of green values of the image;

determine a blue weight using a blue function with the PCA, the blue function comprising a plurality of blue values of the image; and

apply the red weight, the green weight, the blue weight to the channel mixer.

25. The system of claim 24, the processor being further configured to:

determine a yellow weight that is a complement of the red weight;

determine a cyan weight that is a complement of the green weight;

determine a magenta weight that is a complement of the blue weight; and

apply the cyan weight, and the magenta weight to the channel mixer.

26. The system of claim 23, the processor being further configured to:

normalize the principal component to obtain a plurality of normalized components of the principal component; and

apply the plurality of weights based on the plurality of normalized components.

27. The system of claim 23, the processor being further configured to perform a six-way PCA.

28. The system of claim 27, the processor being further configured to:

determine a red value, a green value, and a blue value for a pixel of the image;

arid

determine a red function comprising the red value, a green function comprising the green value, and a blue function comprising the blue value.

29. The system of claim 28, the processor being further configured to:

determine a yellow function comprising a first minimum of the red value and the green value;

determine a cyan function comprising a first minimum of the green value and the blue value;

determine a magenta function comprising a first minimum of the blue value and the red value;

determine the principal component using the red function, the green function, the blue function, the yellow function, the cyan function, and the magenta function; and

wherein the principal component comprises a red component, a green component, and a blue component, a yellow component, a cyan component, and a magenta component.

30. The system of claim 28, the processor being further configured to:

determine a maximum function comprising a maximum value of the red value, the green value, and the blue value;

determine a middle function comprising a middle value of the red value, the green value, and the blue value;

determine a minimum function comprising a minimum value of the red value, the green value, and the blue value;

determine the principal component using the red function, the green function, the blue function, the maximum function, the middle function, and the minimum function; and

wherein the principal component comprises a red component, a green component, and a blue component, a maximum component, a middle component, and a minimum component.

31. The system of claim 30, the processor being further configured to:

determine a red weight comprising a first sum of the red component and the maximum component;

determine a yellow weight comprising a second sum of the red component, the green component, the middle component, and the maximum component;

determine a green weight comprising a third sum of the green component and the maximum component;

determine a cyan weight comprising a fourth sum of the green component, the blue component, the middle component and the maximum component;

determine a blue weight comprising a fifth sum of the blue component and the maximum component;

determine a magenta weight comprising a sixth sum of the blue component, the red component, the middle component, and the maximum component; and

apply the red weight, the green weight, the blue weight, the yellow weight, the cyan weight, and the magenta weight to the channel mixer.

32. A non-transitory computer-readable medium comprising computer instructions for:

performing a principal component analysis (PCA) on an image to determine a principal component;

applying a plurality of weights based on the principal component to a channel mixer; and

generating a grayscale image from the plurality of weights using the channel mixer.

33. The non-transitory computer-readable medium of claim 32, the computer instructions further comprising:

determining a red weight using a red function with the PCA, the red function comprising a plurality of red values of the image;

determining a green weight using a green function with the PCA, the green function comprising a plurality of green values of the image;

determining a blue weight using a blue function with the PCA, the blue function comprising a plurality of blue values of the image; and

applying the red weight, the green weight, the blue weight to the channel mixer.

34. The non-transitory computer-readable medium of claim 33, the computer instructions further comprising:

determining a yellow weight that is a complement of the red weight;

determining a cyan weight that is a complement of the green weight;

determining a magenta weight that is a complement of the blue weight; and

applying the cyan weight, and the magenta weight to the channel mixer.

35. The non-transitory computer-readable medium of claim 32, the computer instructions further comprising:

normalizing the principal component to obtain a plurality of normalized components of the principal component; and

wherein applying the plurality of weights comprises applying the plurality of weights based on the plurality of normalized components.

36. The non-transitory computer-readable medium of claim 32, wherein performing the PCA comprises performing a six-way PCA.

37. The non-transitory computer-readable medium of claim 36, wherein performing the PCA comprises:

determining a red value, a green value, and a blue value for a pixel of the image; and

determining a red function comprising the red value, a green function comprising the green value, and a blue function comprising the blue value.

38. The non-transitory computer-readable medium of claim 37, the computer instructions further comprising:

determining a yellow function comprising a first minimum of the red value and the green value;

determining a cyan function comprising a first minimum of the green value and the blue value;

determining a magenta function comprising a first minimum of the blue value and the red value;

determining the principal component using the red function, the green function, the blue function, the yellow function, the cyan function, and the magenta function; and

wherein the principal component comprises a red component, a green component, and a blue component, a yellow component, a cyan component, and a magenta component.

39. The non-transitory computer-readable medium of claim 37, the computer instructions further comprising:

determining a maximum function comprising a maximum value of the red value, the green value, and the blue value;

determining a middle function comprising a middle value of the red value, the green value, and the blue value;

determining a minimum function comprising a minimum value of the red value, the green value, and the blue value;

determining the principal component using the red function, the green function, the blue function, the maximum function, the middle function, and the minimum function; and

wherein the principal component comprises a red component, a green component, and a blue component, a maximum component, a middle component, and a minimum component.

40. The non-transitory computer-readable medium of claim 39, the computer instructions further comprising:

determining a red weight comprising a first sum of the red component and the maximum component;

determining a yellow weight comprising a second sum of the red component, the green component, the middle component, and the maximum component;

determining a green weight comprising a third sum of the green component and the maximum component;

determining a cyan weight comprising a fourth sum of the green component, the blue component, the middle component and the maximum component;

determining a blue weight comprising a fifth sum of the blue component and the maximum component;

determining a magenta weight comprising a sixth sum of the blue component, the red component, the middle component, and the maximum component; and

applying the red weight, the green weight, the blue weight, the yellow weight, the cyan weight, and the magenta weight to the channel mixer.