Method and system for normalizing a plurality of signals having a shared component

• US 6,687,414 B1
• Filed: 04/18/2000
• Issued: 02/03/2004
• Est. Priority Date: 08/20/1999
• Status: Expired due to Fees

• Current Assignee: Eastman Kodak Company
• Original Assignee: Eastman Kodak Company
• Inventors: Edgar, Albert D.
• Primary Examiner(s): Mehta, Bhavesh M.
• Assistant Examiner(s): Kassa, Yosef
• Application Number:
  US09/551,129
• Time in Patent Office: 1,386 Days
• Field of Search: 382/163, 382/164, 382/165, 382/167, 382/172, 382/190, 382/199, 382/216, 382/218, 382/219, 382/252, 382/251, 382/255, 382/263, 382/260, 382/261, 382/264, 382/266, 382/269, 382/274, 382/275, 382/277, 382/278, 382/282, 382/295, 382/309, 358/447, 358/461, 358/463, 358/518, 358/519, 358/520, 358/521, 358/527, 358/532, 358/537, 345/611, 345/616, 345/617, 345/618, 345/687, 345/694, 348/69, 348/70, 348/743, 348/746, 348/806, 348/807
• US Class Current: 382/275
• CPC Class Codes: G06T 2207/10008 from scanner, fax or copier
• CPC Class Codes: G06T 2207/10024 Color image
• CPC Class Codes: G06T 2207/20032 Median filtering
• CPC Class Codes: G06T 5/20 using local operators
• CPC Class Codes: G06T 5/40 using histogram techniques
• CPC Class Codes: G06T 5/50 using two or more images, e.g. averaging or subtraction
• CPC Class Codes: G06T 5/90 Dynamic range modification of images or parts thereof

Abstract

One aspect of the invention is a method for normalizing a plurality of signals wherein the plurality of signals have a shared component and wherein at least one of the signals has been distorted in a nonlinear way. A distortion function is determined for at least one of the signals which is proportional to the distortion of that signal relative to at least one of the remaining signals. An inverse relative distortion function is generated for the distorted signal responsive to the distortion function that was determined for that signal. The signal is normalized by applying the inverse relative distortion function that was generated for the distorted signal.

Claims

1. A method for normalizing a plurality of signals, wherein each of the plurality of signals have a shared component and wherein at least one of the plurality of signals has been distorted in a nonlinear way, comprising:

determining a distortion function for at least one of the plurality of signals, wherein the distortion function for a particular one signal of the plurality of signals is proportional to the distortion of the particular one signal relative to at least one of the remaining signals in the plurality of signals;

generating an inverse relative distortion function for the at least one of the plurality of signals, wherein the inverse relative distortion function for a particular one signal of the plurality of signals is responsive to the distortion function that was determined for the particular one signal; and

normalizing the at least one of the plurality of signals by applying the inverse relative distortion function that was generated for that one of the plurality of signals.

2. The method of claim 1, further comprising:

filtering the at least one of the plurality of signals prior to determining the distortion function for the at least one of the plurality of signals.

3. The method of claim 2, wherein the at least one of the plurality of signals is filtered using a median filter.

4. The method of claim 1, wherein each of the plurality of signals is a function of a single variable and wherein the distortion function of the at least one of the plurality of signals depends upon the slope of the at least one of the plurality of signals at a plurality of points.

5. The method of claim 1, wherein each of the plurality of signals is a function of multiple variables and wherein the distortion function of the at least one of the plurality of signals depends upon the rate of change of the at least one of the plurality of signals in a plurality of directions at a plurality of points.

6. The method of claim 1, wherein the distortion function of the at least one of the plurality of signals depends upon the gradient of the at least one of the plurality of signals at a plurality of points.

7. The method of claim 1, wherein each of the plurality of signals is a discrete function;

and wherein the distortion function of the at least one of the plurality of signals is derived from a histogram, wherein a first histogram point related to a first point in the plurality of signals is determined by;

computing a first gradient comprising the gradient of the at least one of the plurality of signals at the first point;

computing a second gradient comprising the gradient of at least another one of the plurality of signals at the first point;

computing a first histogram index in response to the magnitude of the first gradient and second gradient;

computing a first weight in response to the cosine of the angle between the first gradient and second gradient and the magnitude of the first gradient and second gradient; and

determining the first histogram point in response to the first histogram index and a second histogram index proportional to the magnitude of the at least one of the plurality of signals at the first point, wherein the weight of the first histogram point is proportional to the first weight.

8. The method of claim 7, further comprising:

fitting a curve to points on the histogram using a curve-fitting algorithm.

9. The method of claim 8, wherein the inverse relative distortion function of the at least one of the plurality of signals depends upon the area under the curve.

10. The method of claim 1, wherein the distortion function for the at least one of the plurality of signals depends upon a histogram comprising a series of points, wherein each point of the histogram depends upon the gradient of the at least one of the plurality of signals at a particular point and the gradient of another one of the plurality of signals at the particular point.

11. The method of claim 10, further comprising:

fitting a curve to points on the histogram using a curve-fitting algorithm.

12. The method of claim 11, wherein the inverse relative distortion function of the at least one of a plurality of channels depends upon the area under the curve.

13. The method of claim 12, further comprising:

filtering, using a median filter, the at least one of the plurality of signals prior to determining the distortion function for the at least one of the plurality of signals.

14. The method of claim 1, further comprising:

determining a distortion function for each of the plurality of signals, wherein the distortion function for a particular one signal of the plurality of signals is proportional to the distortion of the particular one signal relative to at least one of the remaining signals in the plurality of signals; and

generating an inverse relative distortion function for each of the plurality of signals, wherein the inverse relative distortion function for a particular one signal of the plurality of signals is responsive to the distortion function that was determined for the particular one signal.

15. A method of enhancing a digital image, comprising:

determining a distortion function for at least one of a plurality of channels of the digital image, wherein the distortion function for a particular one channel of the plurality of channels is proportional to the distortion of the particular one channel relative to at least one of the remaining channels in the plurality of channels;

generating an inverse relative distortion function for the at least one of the plurality of channels of the digital image, wherein the inverse relative distortion function for a particular one channel of the plurality of channels is responsive to the distortion function that was determined for the particular one channel; and

normalizing the at least one of the plurality of channels by applying the inverse relative distortion function that was generated for that one of the plurality of channels.

16. The method of claim 15, further comprising:

filtering the at least one of the plurality of channels prior to determining the distortion function for the at least one of the plurality of channels.

17. The method of claim 16, wherein the at least one of the plurality of channels is filtered using a median filter.

18. The method of claim 15, wherein the distortion function of the at least one of the plurality of channels depends upon the rate of change of the at least one of the plurality of channels in a plurality of directions at a plurality of points.

19. The method of claim 15, wherein the distortion function of the at least one of the plurality of channels depends upon the gradient of the at least one of the plurality of channels at a plurality of points.

20. The method of claim 19, wherein the at least one of the plurality of channels comprises a plurality of pixel values in a two dimensional plane and wherein the gradient upon which the distortion function depends is computed, for a first pixel, by calculating approximate slopes using other pixel values in the neighborhood of the first pixel.

21. The method of claim 19, wherein the distortion function of the at least one of the plurality of channels further depends upon the gradient of at least another one of the plurality of channels at a plurality of points.

22. The method of claim 15, wherein each of the plurality of channels is a discrete function;

and wherein the distortion function of the at least one of the plurality of channels is derived from a histogram, wherein a first histogram point related to a first point in one of the plurality of channels is determined by;

computing a first gradient comprising the gradient of the at least one of the plurality of channels at the first point;

computing a second gradient comprising the gradient of another one of the plurality of channels at the first point;

computing a first histogram index in response to the magnitude of the first gradient and second gradient;

computing a first weight in response to the cosine of the angle between the first gradient and second gradient and the magnitude of the first gradient and second gradient; and

determining the first histogram point in response to the first histogram index and a second histogram index proportional to the magnitude of the at least one of the plurality of signals at the first point, wherein the weight of the first histogram point is proportional to the first weight.

23. The method of claim 22, further comprising:

fitting a curve to points on the histogram using a curve-fitting algorithm.

24. The method of claim 23, wherein the inverse relative distortion function of the at least one of the plurality of channels depends upon the area under the curve.

25. The method of claim 15, wherein the distortion function for the at least one of the plurality of channels depends upon a histogram comprising a series of points, wherein each point of the histogram depends upon the gradient of the at least one of the plurality of channels at a particular point and the gradient of another on or more of the plurality of channels at the particular point.

26. The method of claim 25, further comprising:

fitting a curve to points on the histogram using a curve-fitting algorithm.

27. The method of claim 26, wherein the inverse relative distortion function of the at least one of the plurality of channels depends upon the area under the curve.

28. The method of claim 27, further comprising:

filtering, using a median filter, the at least one of the plurality of channels prior to determining the distortion function for the at least one of the plurality of channels.

29. The method of claim 15, further comprising:

determining a distortion function for each of the plurality of channels, wherein the distortion function for a particular one channel of the plurality of channels is proportional to the distortion of the particular one channel relative to at least one of the remaining channels in the plurality of channels; and

generating an inverse relative distortion function for each of the plurality of channels, wherein the inverse relative distortion function for a particular one channel of the plurality of channels is responsive to the distortion function that was determined for the particular one channel.

30. The method of claim 29, wherein the distortion function for each of the plurality of channels depends, for a particular channel, upon a histogram comprising a series of points, wherein each point of the histogram depends upon the gradient of the particular channel at a particular point and the gradient of another one or more of the plurality of channels at the particular point.

31. The method of claim 30, further comprising:

fitting a curve to points on each of the histograms using a curve-fitting algorithm.

32. The method of claim 31, wherein the inverse distortion function of each of the plurality of channels depends, for a particular channel upon the area under the curve associated with that particular channel.

33. The method of claim 32, further comprising:

filtering, using a median filter, each of the plurality of channels prior to determining the distortion function for each of the plurality of channels.

34. A digital image scanning system comprising:

scanning hardware operable to scan a photographic image and convert the photographic image into a digital image; and

computer software associated with the scanning hardware and operable to;

determine a distortion function for at least one of a plurality of channels of the digital image, wherein the distortion function for a particular one channel of the plurality of channels is proportional to the distortion of the particular one channel relative to at least one of the remaining channels in the plurality of channels;

generate an inverse relative distortion function for the at least one of the plurality of channels of the digital image, wherein the inverse relative distortion function for the particular one channel of the plurality of channels is responsive to the distortion function that was determined for the particular one channel; and

normalize the at least one of the plurality of channels by applying the inverse relative distortion function that was generated for that one of the plurality of channels.

35. The digital image scanning system of claim 34, wherein the computer software is further operable to:

filter the at least one of the plurality of channels prior to determining the distortion function for the at least one of the plurality of channels.

36. The digital image scanning system of claim 34, wherein the distortion function of the at least one of the plurality of channels depends upon the rate of change of the at least one of the plurality of channels in a plurality of directions at a plurality of points.

37. The digital image scanning system of claim 34, wherein each of the plurality of channels is a discrete function;

and wherein the distortion function of the at least one of the plurality of channels is derived from a histogram by the computer software, wherein a first histogram point related to a first point in one of the plurality of channels is determined by;

computing a first gradient comprising the gradient of the at least one of the plurality of channels at the first point;

computing a second gradient comprising the gradient of another one of the plurality of channels at the first point;

computing a first histogram index in response to the magnitude of the first gradient and second gradient;

computing a first weight in response to the cosine of the angle between the first gradient and second gradient and the magnitude of the first gradient and second gradient; and

determining the first histogram point in response to the first histogram index and a second histogram index proportional to the magnitude of the at least one of the plurality of signals at the first point, wherein the weight of the first histogram point is proportional to the first weight.

38. The digital image scanning system of claim 37, wherein the computer software is further operable to fit a curve to points on the histogram using a curve-fitting algorithm.

39. The digital image scanning system of claim 38, wherein the inverse relative distortion function of the at least one of the plurality of channels depends upon the area under the curve.

40. A digital image processing system comprising:

a computer readable storage medium; and

computer software stored on the computer readable storage medium and operable to;

determine a distortion function for each of a plurality of channels of a digital image, wherein the distortion function for a particular one channel of the plurality of channels is proportional to the distortion of the particular one channel relative to each of the remaining channels in the plurality of channels;

generate an inverse relative distortion function for each of the plurality of channels of the digital image, wherein the inverse relative distortion function for a particular one channel of the plurality of channels is proportional to the distortion function that was determined for the particular one channel; and

normalize each of the plurality of channels by applying, for each one of the plurality of channels, the inverse relative distortion function that was generated for that one of the plurality of channels.

41. The digital image processing system of claim 40, wherein the computer software is further operable to:

filter the at least one of the plurality of channels prior to determining the distortion function for the at least one of the plurality of channels.

42. The digital image processing system of claim 40, wherein the distortion function of the at least one of the plurality of channels depends upon the rate of change of the at least one of the plurality of channels in a plurality of directions at a plurality of points.

43. The digital image processing system of claim 40, wherein each of the plurality of channels is a discrete function;

and wherein the distortion function of the at least one of the plurality of channels is derived from a histogram by the computer software, wherein a first histogram point related to a first point in one of the plurality of channels is determined by;

computing a first gradient comprising the gradient of the at least one of the plurality of channels at the first point;

computing a second gradient comprising the gradient of another one of the plurality of channels at the first point;

computing a first histogram index in response to the magnitude of the first gradient and second gradient;

computing a first weight in response to the cosine of the angle between the first gradient and second gradient and the magnitude of the first gradient and second gradient; and

determining the first histogram point in response to the first histogram index and a second histogram index proportional to the magnitude of the at least one of the plurality of channels at the first point, wherein the weight of the first histogram point is proportional to the first weight.

44. The digital image processing system of claim 43, wherein the computer software is further operable to fit a curve to points on the histogram using a curve-fitting algorithm.

45. The digital image processing system of claim 44, wherein the inverse relative distortion function of the at least one of the plurality of channels depends upon the area under the curve.

46. A method of enhancing a digital image comprising at least two color channels, the method comprising:

generating a first color channel alteration function to apply to a first color channel of the digital image, the first color channel alteration function selected such that when applied to the first color channel, the first color channel and a second color channel of the digital image will vary by approximately equal amounts across small distances in the digital image; and

modifying the first color channel in response to the first color channel alteration function.

47. The method of claim 46, further comprising filtering the first color channel.

48. The method of claim 47, wherein the first color channel alteration function depends upon the gradient of the first color channel at various points after the first color channel has been filtered.

49. The method of claim 46, wherein the digital image comprises a the first color channel, second color channel, and a third color channel, the method further comprising:

generating a second color channel alteration function to apply to the second color channel of the digital image, the second color channel alteration function selected such that when applied to the second color channel, the first color channel, second color channel and third color channel of the digital image will vary by approximately equal amounts across small distances in the digital image;

generating a third color channel alteration function to apply to a third color channel of the digital image, the third color channel alteration function selected such that when applied to the third color channel, the first color channel, second color channel, and third color channel of the digital image will vary by approximately equal amounts across small distances in the digital image;

modifying the second color channel in response to the second color channel alteration function; and

modifying the third color channel in response to the third color channel alteration function, wherein the first color channel alteration function is selected such that when applied to the first color channel, the first color channel, second color channel, and third color channel of the digital image will vary by approximately equal amounts across small distances in the digital image.

50. The method of claim 49, further comprising:

filtering the first color channel, second color channel, and third color channel prior to generating the color channel alteration function associated with each channel.

51. The method of claim 50, wherein the first, second, and third color alteration functions each depend upon the rate of change of the first, second, and third color channels in a plurality of directions at a plurality of points.

52. The method of claim 49, further comprising:

filtering the first color channel;

filtering the second color channel; and

filtering the third color channel, wherein the first, second, and third color alteration functions each depend upon the rate of change of the filtered first, second, and third color channels in a plurality of directions at a plurality of points.

53. An altered digital image derived from a digital image, comprising:

a computer readable storage medium; and

an altered digital image stored on the computer readable storage medium wherein the altered digital image was created by;

generating a first color channel alteration function to apply to a first color channel of the digital image, the first color channel alteration function selected such that when applied to the first color channel, the first color channel and a second color channel of the digital image will vary by approximately equal amounts across small distances in the digital image, and

modifying the first color channel in response to the first color channel alteration function.

54. The altered digital image of claim 53, wherein the altered digital image was further created by filtering the first color channel.

55. The altered digital image of claim 54, wherein the first color channel alteration function depends upon the gradient of the first color channel at various points after the first color channel has been filtered.

56. The method of claim 53, wherein the digital image comprises the first color channel, second color channel, and a third color channel, and wherein the altered digital image was further created by:

generating a second color channel alteration function to apply to the second color channel of the digital image, the second color channel alteration function selected such that when applied to the second color channel, the first color channel, second color channel and third color channel of the digital image will vary by approximately equal amounts across small distances in the digital image;

generating a third color channel alteration function to apply to a third color channel of the digital image, the third color channel alteration function selected such that when applied to the third color channel, the first color channel, second color channel, and third color channel of the digital image will vary by approximately equal amounts across small distances in the digital image;

modifying the second color channel in response to the second color channel alteration function; and

modifying the third color channel in response to the third color channel alteration function, wherein the first color channel alteration function is selected such that when applied to the first color channel, the first color channel, second color channel, and third color channel of the digital image will vary by approximately equal amounts across small distances in the digital image.