System and method for determining a natural color depth of a digital image
View Patent ↗A system and method for determining a natural color depth of a digital image is described. Color distances between each pair of colors in a color palette stored for a digital color image are determined. A closest neighboring color for each color in the color palette separated by a substantially minimum color distance is selected. Each color pair that is visually indistinct is merged in iterative sequence.
1. A system for determining a natural color depth of a digital image, comprising:
a color distance module determining color distances between each pair of colors in a color palette stored for a digital color image;
a merge module selecting a closest neighboring color for each color in the color palette separated by a substantially minimum color distance, and merging, in iterative sequence, the colors in each color pair that are visually indistinct.
2. A system according to claim 1 , further comprising:
an averaging module generating a weighted average of the colors in each color pair that are visually indistinct.
3. A system according to claim 2 , wherein the weighted average z is calculated, in accordance with the equation:
z
=
[
count
(
x
)
×
x
]
+
[
count
(
y
)
×
y
]
count
(
x
)
+
count
(
y
)
;
where x and y are the colors in the color pair, and count( ) denotes the number of occurrences of each color in the digital color image.
4. A system according to claim 1 , wherein visual distinctness is determined based on an output of a function ƒ of visual area, in accordance with the equation:
f
(
x
)
=
c
x
+
c
(
y
2
-
y
1
)
-
1
+
y
1
;
where x is a visual area occupied by one of the colors in the color pair, y 1 denotes a horizontal asymptote, y 2 denotes a y-intercept, and c represents a fitted curve constant.
5. A system according to claim 4 , further comprising:
a visualization module determining visual distinction Δ, in accordance with the equation:
Δ
(
x
,
y
)
=
{
TRUE
,
if
x
-
y
≥
f
(
v
(
x
)
)
;
FALSE
,
if
x
-
y
<
f
(
v
(
x
)
)
;
where v is a visual area comprising a largest four-way block of contiguous pixels of one of the colors in the color pair.
6. A method for determining a natural color depth of a digital image, comprising;
determining color distances between each pair of colors in a color palette stored for a digital color image;
selecting a closest neighboring color for each color in the color palette separated by a substantially minimum color distance; and
merging, in iterative sequence, the colors in each color pair that are visually indistinct.
7. A method according to claim 6 , further comprising:
generating a weighted average of the colors in each color pair that are visually indistinct.
8. A method according to claim 7 , further comprising:
calculating the weighted average z, in accordance with the equation:
z
=
[
count
(
x
)
×
x
]
+
[
count
(
y
)
×
y
]
count
(
x
)
+
count
(
y
)
;
where x and y are the colors in the color pair, and count ( ) denotes the number of occurrences of each color in the digital color image.
9. A method according to claim 6 , further comprising:
determining visual distinctness based on an output of a function ƒ of visual area, in accordance with the equation:
f
(
x
)
=
c
x
+
c
(
y
2
-
y
1
)
-
1
+
y
1
;
where x is a visual area occupied by one of the colors in the color pair, y 1 denotes a horizontal asymptote, Y 2 denotes a y-intercept, and c represents a fitted curve constant.
10. A method according to claim 9 , further comprising:
determining visual distinction Δ, in accordance with the equation:
Δ
(
x
,
y
)
=
{
TRUE
,
if
x
-
y
≥
f
(
v
(
x
)
)
;
FALSE
,
if
x
-
y
<
f
(
v
(
x
)
)
;
where v is a visual area comprising a largest four-way block of contiguous pixels of one of the colors in the color pair.
11. A computer-readable storage medium, holding code for performing the method according to claim 6 , 7 , 8 , 9 , or 10 .
12. A system for reducing a color palette of a digital image to a natural color depth, comprising:
a distance module selecting a neighboring color closest in color distance in a color space for each unique color in the color palette of a digital image; and
a merge module merging the unique color and the closest neighboring color, comprising determining visual distinctness of the unique color based on the visual area occupied by the unique color in the digital image, and combining the unique color and the closest neighboring color when visually indistinct and reducing the color palette by replacing all occurrences of the unique color and the closest neighboring color with the combined color.
13. A system according to claim 12 , wherein selecting the closest neighboring color and merging the unique color and the closest neighboring color is repeated until no further colors in the color palette merge.
14. A system according to claim 12 , further comprising:
a statistical module counting occurrences of the unique color, counting occurrences of the closest neighboring color, and calculating a weighted average of the unique color and the closest neighboring color to combine the unique color and the closest neighboring color.
15. A system according to claim 12 , further comprising:
a visual distinctness module evaluating the visual area occupied by the unique color relative to color distances in the color space from other colors to determine a threshold below which the unique color and any other such color are visually indistinct.
16. A system according to claim 15 , wherein visual distinctness is set as occurring when the color distance between the unique color and the closest neighboring color exceeds the threshold.
17. A system according to claim 15 , wherein the color distance is determined as a Euclidean distance in the color space.
18. A system according to claim 12 , wherein the digital image is received as an output from another color reduction process.
19. A system according to claim 12 , wherein the digital image is forwarded as an input to another color reduction process.
20. A system according to claim 12 , further comprising:
a non-natural color depth color reduction process performing a color reduction of the color palette of the digital image prior reducing the color palette to a natural color depth.
21. A system according to claim 12 , further comprising:
a compression module compressing the digital image following reduction of the color palette to a natural color depth.
22. A system according to claim 12 , wherein the color space is selected from the group comprising CIELAB, RGB, sRGB, YUV, HSV, HSB, and YCbCr.
23. A method for reducing a color palette of a digital image to a natural color depth, comprising:
selecting a neighboring color closest in color distance in a color space for each unique color in the color palette of a digital image; and
merging the unique color and the closest neighboring color, comprising:
determining visual distinctness of the unique color based on the visual area occupied by the unique color in the digital image; and
combining the unique color and the closest neighboring color when visually indistinct and reducing the color palette by replacing all occurrences of the unique color and the closest neighboring color with the combined color.
24. A method according to claim 23 , further comprising:
repeatedly selecting the closest neighboring color and merging the unique color and the closest neighboring color until no further colors in the color palette merge.
25. A method according to claim 23 , further comprising:
counting occurrences of the unique color;
counting occurrences of the closest neighboring color; and
calculating a weighted average of the unique color and the closest neighboring color to combine the unique color and the closest neighboring color.
26. A method according to claim 23 , further comprising:
evaluating the visual area occupied by the unique color relative to color distances in the color space from other colors to determine a threshold below which the unique color and any other such color are visually indistinct.
27. A method according to claim 26 , further comprising:
setting visual distinctness as occurring when the color distance between the unique color and the closest neighboring color exceeds the threshold.
28. A method according to claim 26 , further comprising:
determining the color distance as a Euclidean distance in the color space.
29. A method according to claim 23 , further comprising:
receiving the digital image as an output from another color reduction process.
30. A method according to claim 23 , further comprising:
forwarding the digital image as an input to another color reduction process.
31. A method according to claim 23 , further comprising:
performing a color reduction of the color palette of the digital image prior reducing the color palette to a natural color depth.
32. A method according to claim 23 , further comprising:
compressing the digital image following reduction of the color palette to a natural color depth.
33. A method according to claim 23 , wherein the color space is selected from the group comprising CIELAB, RGB, sRGB, YUV, HSV, HSB, and YCbCr.
34. A computer-readable storage medium holding code for performing the method according to claim 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 , 31 , 32 , or 33 .