Method and system for enhancing digital images
First Claim
1. A method of enhancing a digital image comprising at least one image channel, comprising:
- transforming at least one spatial domain segment of a plurality of spatial domain segments of the at least one image channel from the spatial domain to the frequency domain to create at least one segment transform, wherein each spatial domain segment comprises one of a plurality of spatial domain segments of the at least one image channel;
generating at least one noise correction function in response to the at least one segment transform, wherein the generating further comprises generating a plurality of cross correlation functions wherein a particular one of the plurality of cross correlations depends upon the cross correlation between a first segment transform derived from a first image channel and one or more spatially corresponding segment transforms from one or more image channels other than the first image channel of the digital image;
applying the at least one noise correction function to produce at least one corrected transform wherein the corrected transform depends upon, directly or indirectly, the at least one segment transform; and
transforming the at least one corrected transform from the frequency domain to the spatial domain to create at least one corrected spatial domain segment.
13 Assignments
0 Petitions
Accused Products
Abstract
One aspect of the invention is a method of enhancing a digital image comprising at least one image channel. At least one spatial domain segment of a plurality of spatial domain segments of an image channel is transformed from the spatial domain to the frequency domain to create at least one segment transform. The spatial domain segment comprises one of a plurality of spatial domain segments of at least one image channel. At least one noise correction function is generated in response to the segment transform. The noise correction function may then be applied to produce at least one corrected transform wherein the corrected transform depends upon, directly or indirectly, the segment transform. The corrected transform may be transformed from the frequency domain to the spatial domain to create at least one corrected spatial domain segment.
217 Citations
89 Claims
-
1. A method of enhancing a digital image comprising at least one image channel, comprising:
-
transforming at least one spatial domain segment of a plurality of spatial domain segments of the at least one image channel from the spatial domain to the frequency domain to create at least one segment transform, wherein each spatial domain segment comprises one of a plurality of spatial domain segments of the at least one image channel;
generating at least one noise correction function in response to the at least one segment transform, wherein the generating further comprises generating a plurality of cross correlation functions wherein a particular one of the plurality of cross correlations depends upon the cross correlation between a first segment transform derived from a first image channel and one or more spatially corresponding segment transforms from one or more image channels other than the first image channel of the digital image;
applying the at least one noise correction function to produce at least one corrected transform wherein the corrected transform depends upon, directly or indirectly, the at least one segment transform; and
transforming the at least one corrected transform from the frequency domain to the spatial domain to create at least one corrected spatial domain segment. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 17, 18, 19, 20, 21)
normalizing the magnitude of the noise in the digital image to a constant value over all frequencies prior to the first transforming step.
-
-
3. The method of claim 1, wherein the generating step further comprises generating at least one filtered transform in response to the at least one segment transform.
-
4. The method of claim 3, wherein the generating step further comprises low pass filtering the at least one segment transform to produce the at least one filtered transform.
-
5. The method of claim 3, wherein the generating step further comprises low pass RMS filtering the at least one segment transform to produce the at least one filtered transform.
-
6. The method of claim 1 wherein the generating step further comprises frequency spreading the at least one segment transform to produce at least one spread transform.
-
7. The method of claim 6, wherein the generating step further comprises generating at least one estimate transform that depends upon the at least one spread transform and the at least one segment transform.
-
8. The method of claim 7, wherein the generating step further comprises creating at least one mask function operable to attenuate noise frequencies and accept image frequencies, the mask function generated in response to the at least one estimate transform.
-
9. The method of claim 8, wherein the generating step further comprises applying the at least one mask function to the at least one estimate transform to form the at least one noise correction function.
-
10. The method of claim 1, wherein the at least one noise correction function is applied such that a preselected residual level of noise remains in the at least one corrected spatial domain segment.
-
11. The method of claim 3, wherein the generating step further comprises frequency spreading the at least one filtered transform to produce at least one spread transform.
-
12. The method of claim 11, wherein the generating step further comprises generating at least one estimate transform that depends upon the at least one spread transform and the at least one filtered transform.
-
13. The method of claim 12, wherein the generating step further comprises creating at least one mask function operable to attenuate noise frequencies and accept image frequencies, the at least one mask function generated in response to the at least one estimate transform.
-
14. The method of claim 13, wherein the generating step further comprises applying the at least one mask function to the at least one estimate transform to form the at least one noise correction function.
-
15. The method of claim 14, wherein the at least one noise correction function is applied such that a preselected residual level of noise remains in the at least one corrected spatial domain segment.
-
17. The method of claim 1, wherein the generating step further comprises creating at least one segment estimate of noise compared to signal in the at least one spatial domain segment.
-
18. The method of claim 17, wherein the at least one segment estimate of noise compared to signal in the at least one spatial domain segment depends upon, directly or indirectly, the at least one segment transform;
- and wherein the generating step further comprises attenuating noise frequencies and accepting image frequencies based upon the at least one estimate of noise compared to signal.
-
19. The method of claim 18, wherein the at least one segment estimate of noise compared to signal in the at least one spatial domain segment depends upon at least one filtered segment transform obtained by filtering the at least one segment transform.
-
20. The method of claim 19, wherein the at least one segment estimate of noise compared to signal in the at least one spatial domain segment depends upon a frequency spread version of the at least one filtered segment transform.
-
21. The method of claim 1, wherein one or more of the plurality of spatial domain segments overlap.
-
16. A method of enhancing a digital image comprising at least one image channel, comprising:
-
transforming at least one spatial domain segment of a plurality of spatial domain segments of the at least one image channel from the spatial domain to the frequency domain to create at least one segment transform, wherein each spatial domain segment comprises one of a plurality of spatial domain segments of the at least one image channel;
generating at least one noise correction function in response to the at least one segment transform, wherein the generating further comprises;
generating at least one filtered transform in response to the at least one segment transform;
frequency spreading the at least one filtered transform to produce at least one spread transform;
generating at least one estimate transform that depends upon the at least one spread transform and the at least one filtered transform;
creating at least one mask function operable to attenuate noise frequencies and accept image frequencies, the at least one mask function generated in response to the at least one estimate transform;
applying the at least one mask function to the at least one estimate transform to form the at least one noise correction function; and
filtering at least one cross correlation to produce that at least one filtered transform, wherein the at least one cross correlation depends upon the cross correlation between the at least one segment transform and at least one spatially corresponding segment transform from an image channel of the digital image other than the at least one image channel;
applying the at least one noise correction function to produce at least one corrected transform wherein the corrected transform depends upon, directly or indirectly, the at least one segment transform; and
transforming the at least one corrected transform from the frequency domain to the spatial domain to create at least one corrected spatial domain segment.
-
-
22. A method of enhancing a multicolor digital image comprising a plurality of color channels, comprising:
-
transforming each of a plurality of spatial domain segments of the color channels from the spatial domain to the frequency domain to create a plurality of segment transforms;
generating a plurality of noise correction functions in response to corresponding ones of the plurality of segment transforms, wherein the generating further comprises generating a plurality of cross correlation functions wherein a particular one of the plurality of cross correlations depends upon the cross correlation between a first segment transform derived from a first color channel and one or more spatially corresponding segment transforms from one or more color channels other than the first color channel of the multicolor digital image;
applying ones of the plurality of noise correction functions to produce a plurality of corrected transforms wherein each corrected transform depends upon, directly or indirectly, a corresponding segment transform; and
transforming ones of the plurality of corrected transforms from the frequency domain to the spatial domain to create a plurality of corrected spatial domain segments. - View Dependent Claims (23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41)
-
-
42. A method of enhancing a multicolor digital image comprising a plurality of color channels, comprising:
-
transforming each of a plurality of spatial domain segments of the color channels from the spatial domain to the frequency domain to create a plurality of segment transforms;
generating a plurality of noise correction functions in response to corresponding ones of the plurality of segment transforms, wherein the generating step further comprises;
frequency spreading ones of the plurality of segment transforms to produce a plurality of spread transforms;
generating a plurality of estimate transforms each dependent upon a corresponding one of the plurality of spread transforms and a corresponding one of the plurality of segment transforms;
creating a plurality of mask functions operable to attenuate noise frequencies and accept image frequencies, each mask function generated in response to a corresponding one of the plurality of estimate transforms;
applying ones of the plurality of mask functions to corresponding ones of the plurality of estimate transforms to form a plurality of noise correction functions; and
filtering at least one cross correlation to produce the at least one filtered transform, wherein the at least one cross correlation depends upon the cross correlation between at least one of the plurality of segment transforms and at least one spatially corresponding one of the plurality of segment transforms from a color channel of the digital image other than the color channel associated with the at least one of the plurality of segment transforms;
applying ones of the plurality of noise correction functions to produce a plurality of corrected transforms wherein each corrected transform depends upon, directly or indirectly, a corresponding segment transform; and
transforming ones of the plurality of corrected transforms from the frequency domain to the spatial domain to create a plurality of corrected spatial domain segments.
-
-
43. 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;
transform at least one spatial domain segment of a plurality of spatial domain segments of the at least one image channel from the spatial domain to the frequency domain to create at least one segment transform, wherein each spatial domain segment comprises one of a plurality of spatial domain segments of the at least one image channel;
generate at least one noise correction function in response to the at least one segment transform, wherein the generating further comprises generating a plurality of cross correlation functions wherein a particular one of the plurality of cross correlations depends upon the cross correlation between a first segment transform derived from a first image channel and one or more spatially corresponding segment transforms from one or more image channels other than the first image channel of the digital image;
apply the at least one noise correction function to produce at least one corrected transform wherein the corrected transform depends upon, directly or indirectly, the at least one segment transform; and
transform the at least one corrected transform from the frequency domain to the spatial domain to create at least one corrected spatial domain segment. - View Dependent Claims (44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 59, 60, 61, 62, 63)
normalize the magnitude of the noise in the digital image to a constant value over all frequencies prior to the first transforming step.
-
-
45. The digital image scanning system of claim 43, wherein the computer software, in connection with the generation of at least one noise correction function is further operable to generate at least one filtered transform in response to the at least one segment transform.
-
46. The digital image scanning system of claim 45 wherein the computer software, in connection with the generation of at least one noise correction function is further operable to low pass filter the at least one segment transform to produce the at least one filtered transform.
-
47. The digital image scanning system of claim 45 wherein the computer software, in connection with the generation of at least one noise correction function is further operable to low pass RMS filter the at least one segment transform to produce the at least one filtered transform.
-
48. The digital image scanning system of claim 43 wherein the computer software, in connection with the generation of at least one noise correction function is further operable to frequency spread the at least one segment transform to produce at least one spread transform.
-
49. The digital image scanning system of claim 48, wherein the computer software, in connection with the generation of at least one noise correction function is further operable to generate at least one estimate transform that depends upon the at least one spread transform and the at least one segment transform.
-
50. The digital image scanning system of claim 49 wherein the computer software, in connection with the generation of at least one noise correction function is further operable to create at least one mask function operable to attenuate noise frequencies and accept image frequencies, the mask function generated in response to the at least one estimate transform.
-
51. The digital image scanning system of claim 50 wherein the computer software, in connection with the generation of at least one noise correction function is further operable to apply the at least one mask function to the at least one estimate transform to form the at least one noise correction function.
-
52. The digital image scanning system of claim 43, wherein the computer software is operable to apply the at least one noise correction function such that a preselected residual level of noise remains in the at least one corrected spatial domain segment.
-
53. The digital image scanning system of claim 45 wherein the computer software, in connection with the generation of at least one noise correction function is further operable to frequency spread the at least one filtered transform to produce at least one spread transform.
-
54. The digital image scanning system of claim 53 wherein the computer software, in connection with the generation of at least one noise correction function is further operable to generate at least one estimate transform that depends upon the at least one spread transform and the at least one filtered transform.
-
55. The digital image scanning system of claim 54 wherein the computer software, in connection with the generation of at least one noise correction function is further operable to create at least one mask function operable to attenuate noise frequencies and accept image frequencies, the at least one mask function generated in response to the at least one estimate transform.
-
56. The digital image scanning system of claim 55, wherein the computer software, in connection with the generation of at least one noise correction function is further operable to apply the at least one mask function to the at least one estimate transform to form the at least one noise correction function.
-
57. The digital image scanning system of claim 56 wherein the computer software is operable to apply the at least one noise correction function such that a preselected residual level of noise remains in the at least one corrected spatial domain segment.
-
59. The digital image scanning system of claim 43, wherein the computer software, in connection with the generation of at least one noise correction function is further operable to create at least one segment estimate of noise compared to signal in the at least one spatial domain segment.
-
60. The digital image scanning system of claim 59 wherein the at least one segment estimate of noise compared to signal in the at least one spatial domain segment depends upon, directly or indirectly, the at least one segment transform;
- and wherein the computer software, in connection with the generation of at least one noise correction function is further operable to attenuate noise frequencies and accept image frequencies based upon the at least one estimate of noise compared to signal.
-
61. The digital image scanning system of claim 60 wherein the at least one segment estimate of noise compared to signal in the at least one spatial domain segment depends upon at least one filtered segment transform obtained by filtering the at least one segment transform.
-
62. The digital image scanning system of claim 61 wherein the at least one segment estimate of noise compared to signal in the at least one spatial domain segment depends upon a frequency spread version of the at least one filtered segment transform.
-
63. The digital image scanning system of claim 43 wherein one or more of the plurality of spatial domain segments overlap.
-
58. 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;
transform at least one spatial domain segment of a plurality of spatial domain segments of the at least one image channel from the spatial domain to the frequency domain to create at least one segment transform, wherein each spatial domain segment comprises one of a plurality of spatial domain segments of the at least one image channel;
generate at least one noise correction function in response to the at least one segment transform, wherein the computer software, in connection with the generation of at least one noise correction function is further operable to;
generate at least one filtered transform in response to the at least one segment transform;
frequency spread the at least one filtered transform to produce at least one spread transform;
generate at least one estimate transform that depends upon the at least one spread transform and the at least one filtered transform;
create at least one mask function operable to attenuate noise frequencies and accept image frequencies, the at least one mask function generated in response to the at least one estimate transform;
apply the at least one mask function to the at least one estimate transform to form the at least one noise correction function, wherein the computer software is operable to apply the at least one noise correction function such that a preselected residual level of noise remains in the at least one corrected spatial domain segment; and
filter at least one cross correlation to produce the at least one filtered transform, wherein the at least one cross correlation depends upon the cross correlation between the at least one segment transform and at least one spatially corresponding segment transform from an image channel of the digital image other than the at least one image channel;
apply the at least one noise correction function to produce at least one corrected transform wherein the corrected transform depends upon, directly or indirectly, the at least one segment transform; and
transform the at least one corrected transform from the frequency domain to the spatial domain to create at least one corrected spatial domain segment.
-
-
64. A digital image processing system comprising:
-
a computer readable storage medium; and
computer software stored on the computer readable storage medium and operable to;
transform at least one spatial domain segment of a plurality of spatial domain segments of the at least one image channel from the spatial domain to the frequency domain to create at least one segment transform, wherein each spatial domain segment comprises one of a plurality of spatial domain segments of the at least one image channel;
generate at least one noise correction function in response to the at least one segment transform, wherein the generating further comprises generating a plurality of cross correlation functions wherein a particular one of the plurality of cross correlations depends upon the cross correlation between a first segment transform derived from a first image channel and one or more spatially corresponding segment transforms from one or more image channels other than the first image channel of a digital image;
apply the at least one noise correction function to produce at least one corrected transform wherein the corrected transform depends upon, directly or indirectly, the at least one segment transform; and
transform the at least one corrected transform from the frequency domain to the spatial domain to create at least one corrected spatial domain segment. - View Dependent Claims (65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 80, 81, 82, 83, 84)
normalize the magnitude of the noise in the digital image to a constant value over all frequencies prior to the first transforming step.
-
-
66. The digital image processing system of claim 64 wherein the computer software, in connection with the generation of at least one noise correction function is further operable to generate at least one filtered transform in response to the at least one segment transform.
-
67. The digital image processing system of claim 66 wherein the computer software, in connection with the generation of at least one noise correction function is further operable to low pass filter the at least one segment transform to produce the at least one filtered transform.
-
68. The digital image processing system of claim 66 wherein the computer software, in connection with the generation of at least one noise correction function is further operable to low pass RMS filter the at least one segment transform to produce the at least one filtered transform.
-
69. The digital image processing system of claim 64, wherein the computer software, in connection with the generation of at least one noise correction function is further operable to frequency spread the at least one segment transform to produce at least one spread transform.
-
70. The digital image processing system of claim 69, wherein the computer software, in connection with the generation of at least one noise correction function is further operable to generate at least one estimate transform that depends upon the at least one spread transform and the at least one segment transform.
-
71. The digital image processing system of claim 70 wherein the computer software, in connection with the generation of at least one noise correction function is further operable to create at least one mask function operable to attenuate noise frequencies and accept image frequencies, the mask function generated in response to the at least one estimate transform.
-
72. The digital image processing system of claim 71, wherein the computer software, in connection with the generation of at least one noise correction function is further operable to apply the at least one mask function to the at least one estimate transform to form the at least one noise correction function.
-
73. The digital image processing system of claim 64, wherein the computer software is operable to apply the at least one noise correction function such that a preselected residual level of noise remains in the at least one corrected spatial domain segment.
-
74. The digital image processing system of claim 66, wherein the computer software, in connection with the generation of at least one noise correction function is further operable to frequency spread that at least one filtered transform to produce at least one spread transform.
-
75. The digital image processing system of claim 74, wherein the computer software, in connection with the generation of at least one noise correction function is further operable to generate at least one estimate transform that depends upon the at least one spread transform and the at least one filtered transform.
-
76. The digital image processing system of claim 75, wherein the computer software, in connection with the generation of at least one noise correction function is further operable to create at least one mask function operable to attenuate noise frequencies and accept image frequencies, the at least one mask function generated in response to the at least one estimate transform.
-
77. The digital image processing system of claim 76, wherein the computer software, in connection with the generation of at least one noise correction function is further operable to apply the at least one mask function to the at least one estimate transform to form the at least one noise correction function.
-
78. The digital image processing system of claim 77, wherein the computer software is operable to apply the at least one noise correction function such that a preselected residual level of noise remains in the at least one corrected spatial domain segment.
-
80. The digital image processing system of claim 64, wherein the computer software, in connection with the generation of at least one noise correction function is further operable to create at least one segment estimate of noise compared to signal in the at least one spatial domain segment.
-
81. The digital image processing system of claim 80, wherein the at least one segment estimate of noise compared to signal in the at least one spatial domain segment depends upon, directly or indirectly, the at least one segment transform;
- and wherein the computer software, in connection with the generation of at least one noise correction function is further operable to attenuate noise frequencies and accept image frequencies based upon the at least one estimate of noise compared to signal.
-
82. The digital image processing system of claim 81, wherein the at least one segment estimate of noise compared to signal in the at least one spatial domain segment depends upon at least one filtered segment transform obtained by filtering the at least one segment transform.
-
83. The digital image processing system of claim 82, wherein the at least one segment estimate of noise compared to signal in the at least one spatial domain segment depends upon a frequency spread version of the at least one filtered segment transform.
-
84. The digital image processing system of claim 64, wherein one or more of the plurality of spatial domain segments overlap.
-
79. A digital image processing system comprising:
-
a computer readable storage medium; and
computer software stored on the computer readable storage medium and operable to;
transform at least one spatial domain segment of a plurality of spatial domain segments of the at least one image channel from the spatial domain to the frequency domain to create at least one segment transform, wherein each spatial domain segment comprises one of a plurality of spatial domain segments of the at least one image channel;
generate at least one noise correction function in response to the at least one segment transform, wherein the computer software, in connection with the generation of at least one noise correction function is further operable to;
generate at least one filtered transform in response to the at least one segment transform, frequency spread that at least one filtered transform to produce at least one spread transform;
generate at least one estimate transform that depends upon the at least one spread transform and the at least one filtered transform;
create at least one mask function operable to attenuate noise frequencies and accept image frequencies, the at least one mask function generated in response to the at least one estimate transform;
apply the at least one mask function to the at least one estimate transform to form the at least one noise correction function, wherein the computer software is operable to apply the at least one noise correction function such that a preselected residual level of noise remains in the at least one corrected spatial domain segment; and
filter at least one cross correlation to produce the at least one filtered transform, wherein the at least one cross correlation depends upon the cross correlation between the at least one segment transform and at least one spatially corresponding segment transform from an image channel of the digital image other than the at least one image channel;
apply the at least one noise correction function to produce at least one corrected transform wherein the corrected transform depends upon, directly or indirectly, the at least one segment transform; and
transform the at least one corrected transform from the frequency domain to the spatial domain to create at least one corrected spatial domain segment.
-
-
85. An altered digital image derived from a digital image comprising at least one image channel, 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;
transforming at least one spatial domain segment of a plurality of spatial domain segments of the at least one image channel from the spatial domain to the frequency domain to create at least one segment transform, wherein each spatial domain segment comprises one of a plurality of spatial domain segments of the at least one image channel;
generating at least one noise correction function in response to the at least one segment transform, wherein the generating further comprises generating a plurality of cross correlation functions wherein a particular one of the plurality of cross correlations depends upon the cross correlation between a first segment transform derived from a first image channel and one or more spatially corresponding segment transforms from one or more image channels other than the first image channel of the digital image;
applying the at least one noise correction function to produce at least one corrected transform wherein the corrected transform depends upon, directly or indirectly, the at least one segment transform; and
transforming the at least one corrected transform from the frequency domain to the spatial domain to create at least one corrected spatial domain segment. - View Dependent Claims (86, 87, 88, 89)
-
Specification