Determining similarity of two images
A computer implemented method is disclosed. The method includes obtaining a first and second images for comparison, globally registering the first and second images, calculating for pairs of a first patch of the first image and second patch of the second image a similarity measure which is a product of luminance and contrast components and a normalized structure component, each component taking a value between 0 and 1 and determining similarity of the images based on the calculated similarity measure. Relating computer program product and data processing system are also disclosed.
1. A computer implemented method comprising:
obtaining, by a processor, a first and second images for comparison;
globally registering, by a processor, the first and second images;
calculating, by a processor, for pairs of a first patch x of the first image and second patch y of the second image a similarity measure SM(x,y) wherein the similarity measure SM satisfies for any
a∈[ 0,1 ],SM ( x,y )≦ SM ( x,ax +(1 −a ) y );
determining similarity measures for each of the pairs of patches; and
determining a similarity of the first and second images based on the calculated similarity measures for each of the pairs of patches.
2. A method as claimed in claim 1 , wherein calculating of the similarity measure includes using the equation
SSI
M
~
(
x
,
y
)
=
[
l
(
x
,
y
)
]
α
·
[
c
(
x
,
y
)
]
β
·
[
1
2
(
s
(
x
,
y
)
+
1
)
]
γ
,
where x and y are the two images, SSI{tilde over (M)}(x, y) is the similarity measure, l(x,y) is the luminance component, c(x,y) is the contrast component, s(x,y) is a structure component and
1
2
(
s
(
x
,
y
)
+
1
)
is the normalized structure component, and where α, β and γ are parameters used to adjust the relative importance of the three components, α>0, β>0, γ>0.
3. A method as claimed in claim 1 , comprising ignoring sub-pixel misregistration in the registration of the two images.
4. A method as claimed in claim 3 , further including calculating, for pairs of corresponding patches of the two images, a further similarity measure SM sp ignoring sub-pixel misregistration given by
SM
sp
(
x
,
y
)
=
min
(
1
,
SM
(
x
,
y
)
G
SM
(
x
)
)
where x and y are the two patches images, SM(x,y) is the calculated similarity measure and where
G
SM
(
x
)
=
min
x
~
∈
N
x
[
δ
·
SM
(
x
,
x
~
)
+
ɛ
]
,
N x being a set of n patches in the n-pixel neighborhood of a pixel in patch x, and where δ and ∈ are constants.
5. A method as claimed in claim 1 , wherein obtaining of the images includes capturing at least one of the images by an image capturing device.
6. A method as claimed in claim 5 , wherein said captured at least one of the images is a printed image.
7. A method as claimed in claim 1 , comprising producing a similarity map.
8. A computer program product stored on a non-transitory tangible computer readable storage medium, the computer program including code for:
obtaining a first and second images for comparison;
globally registering the first and second images;
calculating for pairs of a first patch x of the first image and second patch y of the second image a similarity measure SM(x,y) wherein the similarity measure SM satisfies for any
a∈[ 0,1 ],SM ( x,y )≦ SM ( x,ax +(1 −a ) y );
determining similarity measures for each of the pairs of patches; and
determining a similarity of the first and second images based on the calculated similarity measures for each of the pairs of patches.
9. A computer program product as claimed in claim 8 , including code for calculating of the similarity measure includes using the equation
SSI
M
~
(
x
,
y
)
=
[
l
(
x
,
y
)
]
α
·
[
c
(
x
,
y
)
]
β
·
[
1
2
(
s
(
x
,
y
)
+
1
)
]
γ
,
where x and y are the two images, SSI{tilde over (M)}(x, y) is the similarity measure, l(x,y) is the luminance component, c(x,y) is the contrast component, s(x,y) is a structure component and
1
2
(
s
(
x
,
y
)
+
1
)
is the normalized structure component, and where α, β and γ are parameters used to adjust the relative importance of the three components, α>0, β>0, γ>0.
10. A computer program product as claimed in claim 8 , including code for ignoring sub-pixel misregistration in the registration of the two images.
11. A computer program product as claimed in claim 10 , including code for calculating, for pairs of corresponding patches of the two images, a further similarity measure SM sp ignoring sub-pixel misregistration given by
SM
sp
(
x
,
y
)
=
min
(
1
,
SM
(
x
,
y
)
G
SM
(
x
)
)
where x and y are the two patches images, SM(x,y) is the calculated similarity measure and where
G
SM
(
x
)
=
min
x
~
∈
N
x
[
δ
·
SM
(
x
,
x
~
)
+
ɛ
]
,
N x being a set of n patches in the n-pixel neighborhood of a pixel in patch x, and where δ and ∈ are constants.
12. A computer program product as claimed in claim 8 , including code for capturing at least one of the images by an image capturing device.
13. A computer program product as claimed in claim 12 , including code for printing at least one of the images and wherein said captured at least one of the images is said printed at least one of the images.
14. A computer program product as claimed in claim 8 , comprising producing a similarity map.
15. A data processing system comprising:
a processing unit in communication with a computer usable medium, wherein the computer usable medium contains a set of instructions wherein the processing unit is designed to carry out the set of instructions to:
obtain a first and second images for comparison;
globally register the first and second images;
calculate for pairs of a first patch x of the first image and second patch y of the second image a similarity measure SM(x,y) wherein the similarity measure SM satisfies for any
a∈[ 0,1 ],SM ( x,y )≦ SM ( x,ax +(1 −a ) y );
determining similarity measures for each of the pairs of patches; and
determining a similarity of the first and second images based on the calculated similarity measures for each of the pairs of patches.
16. A system as claimed in claim 15 , wherein the processing unit is designed to carry out the set of instructions to calculate the similarity measure using the equation
SSI
M
~
(
x
,
y
)
=
[
l
(
x
,
y
)
]
α
·
[
c
(
x
,
y
)
]
β
·
[
1
2
(
s
(
x
,
y
)
+
1
)
]
γ
,
where x and y are the two images, SSI{tilde over (M)}(x, y) is the similarity measure, l(x,y) is the luminance component, c(x,y) is the contrast component, s(x,y) is a structure component and
1
2
(
s
(
x
,
y
)
+
1
)
is the normalized structure component, and where α, β and γ are parameters used to adjust the relative importance of the three components, α>0, β>0, γ>0.
17. A system as claimed in claim 15 , wherein the processing unit is designed to carry out the set of instructions to ignore sub-pixel misregistration in the registration of the two images.
18. A system as claimed in claim 17 , wherein the processing unit is designed to carry out the set of instructions to calculate, for pairs of corresponding patches of the two images, a further similarity measure SM sp ignoring sub-pixel misregistration given by
SM
sp
(
x
,
y
)
=
min
(
1
,
SM
(
x
,
y
)
G
SM
(
x
)
)
where x and y are the two patches images, SM(x,y) is the calculated similarity measure and where
G
SM
(
x
)
=
min
x
~
∈
N
x
[
δ
·
SM
(
x
,
x
~
)
+
ɛ
]
,
N x being a set of n patches in the n-pixel neighborhood of a pixel in patch x, and where δ and ∈ are constants.
19. A system as claimed in claim 15 , wherein the processing unit is designed to carry out the set of instructions to capture at least one of the images by an image capturing device.
20. A system as claimed in claim 15 , wherein the processing unit is designed to carry out the set of instructions to print at least one of the images and wherein said captured at least one of the images is said printed at least one of the images.