Gamut mapping preserving local luminance differences with adaptive spatial filtering
View Patent ↗A color printing process, printing a color image in which out-of-gamut original colors are present. For each pixel defined by an original color which is determined to be out of gamut, a gamut remapping process is applied to map each pixel to a color which is within a printer gamut, remapping said pixels to colors within an output printer gamut. For a given set of gamut remapped pixels, gamut remapped pixel colors are compared with said original pixel colors, to derive a comparison metric. Using the comparison metric, a corrected set of gamut remapped colors is generated. The comparison metric may be subjected to an adaptive filtering process, which strengthens the comparison metric in high frequency image regions to increase its impact on the gamut remapped colors, and weakens the comparison metric in low frequency areas, to weaken its impact on the gamut remapped colors. The filter is selected by determining a filter selection metric in accordance with measured local image activity and changing filter parameters as a function of the determined filter selection metric.
1. A method of selecting a filter for control of a gamut mapping correction process, including:
determining a filter selection metric in accordance with measured local image activity;
changing filter parameters as a function of the determined filter selection metric; and,
using the changed filter parameters so that luminance feedback in a gamut mapping correction process is in accordance with local image activity.
2. A method as described in claim 1 , wherein said local image activity metric varies between low activity, corresponding to flat areas within an image, and high activity, corresponding to strong edge areas with an image.
3. A method as described in claim 2 , wherein said image activity corresponds to a data norm of order p, given by
Activity
=
[
∑
j
(
e
i
p
-
e
j
p
)
]
1
ρ
.
4. A method as described in claim 2 , wherein said image activity corresponds to
1
m
∑
j
(
e
i
-
e
j
)
.
wherein e i is a luminance error at a target pixel i and e j is a luminance error at a pixel j within a neighborhood of pixel i, and m is the number of pixels in the neighborhood.
5. A method as derived in claim 1 wherein said varied filter parameter is filter size in terms of pixels covered in a single operation thereof.
6. A method as derived in claim 1 , wherein the filter selection metric and fitter parameters are derived as follows:
computing said activity metric within a small pixel neighborhood neighborhood defined as N s ×N 5
for activity metric values within a predetermined range of activity values, employing a relatively small filter size S 1 ×S 1
for activity metric values outside said predetermined range of activity, computing a activity metric a i over a large pixel neighborhood N l ×N l
compute the ratio R=a i /a s
If R is greater than a predetermined threshold, employ a small filter size S 1
If R is less than the said threshold, employ a large filter size S 2 ×S 2 .
7. A method as in claim 6 wherein N s =5, N l =15, S 1 =5, and S 2 =15.
8. A method of selecting a filter for control of a gamut mapping correction process, comprising:
determining a filter selection metric in accordance with measured local image activity, wherein said local image activity metric varies between low activity, corresponding to flat areas within an image, and high activity, corresponding to strong edge areas with an image, said image activity corresponding to a data norm of order p, given by
Activity
=
[
∑
j
(
e
i
p
-
e
j
p
)
]
1
ρ
;
and,
changing filter parameters as a function of the determined filter selection metric.
9. A method of selecting a filter for control of a gamut mapping correction process, comprising:
determining a filter selection metric in accordance with measured local image activity, wherein said local image activity metric varies between low activity, corresponding to flat areas within an image, and high activity, corresponding to strong edge areas with an image, said image activity corresponds to
1
m
∑
j
(
e
i
-
e
j
)
wherein e i is a luminance error at a target pixel i and e j is a luminance error at a pixel j within a neighborhood of pixel i, and m is the number of pixels in the neighborhood; and,
changing filter parameters as a function of the determined filter selection metric.
10. A method of selecting a fitter for control of a gamut mapping correction process, comprising:
determining a filter selection metric in accordance with measured local image activity metric; and,
changing filter parameters as a function of the determined filter selection metric, the filter selection metric and filter parameters being derived as follows:
computing said image activity metric within a small pixel neighborhood, the neighborhood defined as N s ×N s
for image activity metric values within a predetermined range of activity values, employing a relatively small filter size S 1 ×S 1
for image activity metric values outside said predetermined range of activity, computing an activity metric a L over a large pixel neighborhood N l ×N l
compute the ratio R=a L /a s
if R is greater than a predetermined threshold, employ a small filter size S 1
if R is less than the said threshold, employ a large filter size S 2 ×S 2 .
11. A method as in claim 10 wherein Ns=5, Nl=15, S1=5, and S2=15.