Method and apparatus for automatically determining image foreground color
View Patent ↗A foreground color for a digital image is automatically determined by dividing the colors of the pixels of at least a part of the digital image into a number of color clusters in a color space, and for at least one cluster selecting a color being related to the at least one color cluster according to predetermined criteria.
1. A method for automatically determining a foreground color for a digital image, comprising:
(a) automatically dividing the colors of the pixels of at least a part of the digital image into a number of color clusters in a color space;
(b) automatically segmenting the part of the digital image into regions according to the color clusters;
(c) automatically grouping the color clusters into interference clusters comprising a pixel in a selected region of the image and benign clusters comprising no pixel in the selected region;
(d) automatically selecting a foreground color being related to at least one interference cluster according to predetermined criteria and the foreground color being selected according to a color harmony criterion with respect to at least one benign cluster; and
(e) one of automatically displaying, storing and communicating data representing the selected foreground color.
2. The method according to claim 1 , wherein said selecting at (d) further comprises:
selecting a harmonious color set with respect to the color clusters; and
testing the harmonious color set for legibility.
3. The method according to claim 2 , wherein said testing the harmonious color set for legibility further comprises:
computing local measures of contrast between a background and the foreground in a neighborhood of a predetermined foreground region; and
computing a legibility score representative of a lowest few contrast values observed over the predetermined foreground region.
4. The method according to claim 2 , wherein the foreground color is selected according to a legibility criterion for a predetermined foreground region by:
computing local measures of contrast between background and foreground in a neighborhood for a predetermined foreground region; and
computing a legibility score representative of a lowest few contrast values observed over the predetermined foreground region.
5. The method according to claim 1 , wherein said dividing at (a) comprises converting the image data to a predetermined color format.
6. The method according to claim 1 , wherein said dividing at (a) comprises using an Expectation-Maximization clustering.
7. The method according to claim 1 , wherein said dividing at (a) comprises determining the number of clusters using a model selection method one of a Bayesian Information Criterion and a Universal Model-based Minimum Description Length Principle.
8. The method according to claim 1 , wherein the segmenting comprises using one of a normalized cut criterion and an energy-minimization method.
9. The method according to claim 1 , wherein the foreground color is selected according to a legibility criterion for a predetermined foreground region.
10. The method according to claim 9 , wherein the foreground color is selected based on a likelihood ratio of the hypothesis that the digital image contains the foreground region and the hypothesis that the digital image does not contain the foreground region.
11. The method according claim 10 , wherein selecting the foreground color comprises computing the legibility of
min
x
∈
T
max
y
∈
C
ɛ
2
r
(
x
+
y
)
,
wherein
r
(
x
)
=
h
log
Pr
(
I
(
x
)
|
T
)
Pr
(
I
(
x
)
|
B
)
,
C ε 2 is a disc of radius ε and wherein Pr(i(x)|T) denotes heuristic or other models of likelihoods that the image I contains text T at a given pixel x and Pr(J(x)|B) denotes heuristic or other models of likelihoods that the image I contains background B at the given pixel x.
12. The method according to claim 1 , wherein the foreground color is selected according to at least one of a monotonic, a complementary, and a p-adic color harmony criterion in HSL space.
13. The method according to claim 1 , wherein the foreground color is selected according to a color harmony criterion with respect to at least one interference cluster.
14. The method according to claim 1 , wherein the foreground color is selected according to a color harmony criterion with respect to all interference clusters and at least one benign cluster.
15. The method according to claim 1 , wherein said selecting at (d) comprises determining a color subset according to a color harmony criterion and maximizing a legibility function in the color subset.
16. The method according to claim 1 , wherein the foreground color is selected for which
∑
i
=
1
M
α
1
l
(
c
,
P
i
)
+
∑
k
=
1
N
γ
k
h
(
c
,
K
k
)
is maximal, wherein c denotes the foreground color, P i denote the interference clusters, K k denote all clusters, both benign and interference, I is a legibility function in color space, h is a color harmony function, and α i and γ k are weighting factors.
17. The method according to claim 1 , further comprising one of displaying and storing a predetermined object using the selected foreground color together with the digital image.
18. An apparatus for automatically determining a foreground color for a digital image comprising:
a processor which executes computer readable instructions, stored in a memory, to perform the method comprising:
(a) automatically dividing the colors of the pixels of at least a part of the digital image into a number of color clusters in a color space;
(b) automatically segmenting the part of the digital image into regions according to the color clusters;
(c) automatically grouping the color clusters into interference clusters comprising a pixel in a selected region of the image and benign clusters comprising no pixel in the selected region;
(d) automatically selecting a foreground color being related to at least one interference cluster according to predetermined criteria and the foreground color being selected according to a color harmony criterion with respect to at least one benign cluster; and
(e) one of automatically displaying, storing and communicating data representing the selected foreground color.
19. The apparatus according to claim 18 , wherein said selecting at (d) further comprises:
selecting a harmonious color set with respect to the color clusters; and
testing the harmonious color set for legibility.
20. The apparatus according to claim 19 , wherein said testing the harmonious color set for legibility further comprises:
computing local measures of contrast between background and foreground in a neighborhood of a predetermined foreground region; and
computing a legibility score representative of a lowest few contrast values observed over the predetermined foreground region.
21. A computer program product comprising:
a computer-usable memory device storing instructions that, when executed by a computer, cause the computer to perform a method comprising:
(a) automatically dividing the colors of the pixels of at least a part of the digital image into a number of color clusters in a color space;
(b) automatically segmenting the part of the digital image into regions according to the color clusters;
(c) automatically grouping the color clusters into interference clusters comprising a pixel in a selected region of the image and benign clusters comprising no pixel in the selected region;
(d) automatically selecting a foreground color being related to at least one interference cluster according to predetermined criteria and the foreground color being selected according to a color harmony criterion with respect to at least one benign cluster; and
(e) one of automatically displaying, storing and communicating data representing the selected foreground color.
22. The computer program product according to claim 21 , wherein said selecting at (d) further comprises:
selecting a harmonious color set with respect to the color clusters; and
testing the harmonious color set for legibility.
23. The computer program product according to claim 22 , wherein said testing the harmonious color set for legibility further comprises:
computing local measures of contrast between background and foreground in a neighborhood of a predetermined foreground region; and
computing a legibility score representative of a lowest few contrast values observed over the predetermined foreground region.