Image Processing Reference
In-Depth Information
sRGB
in
ROMMRGB
in
Lab
in
Gamut
mapping
T
1
T
2
Mapped
Lab
CMY
in
ˆ
P
-1
CMYK
printer
P
UCR/GCR
P
1
Compute
P
-1
using ICI
ˆ
Lab
out
Downsample
P
using dynamic
optimization (DO)
~
Upsample
~
using trilinear
interpolation
For visualization on a typical display screen
sRGB
out
ROMMRGB
out
T
1
-1
T
2
-1
FIGURE 6.21
Printing system image path.
6.6 SMOOTHING ALGORITHM FOR MULTIDIMENSIONAL
FUNCTIONS
6.6.1 I
NTRODUCTION
Data
fitting is an important task in printer calibration and characterization. When
fitting a function to data, there are two important considerations. The
first is the
goodness of the
fit. Ideally, the
fit and the data would have the same value at every
point. The goodness of the
fit is typically measured by an error metric de
ned in
terms of the difference between the available data and the
t. The L
2
norm (MSE) is a
measure normally used for this purpose. Other norms such as L
1
and L
1
can also be
used (Section 3.10.1). The second consideration is the smoothness of the
t. The
smoothness of the
first or second derivative of
the underlying function. If the function is smooth, then its second derivative must
fit is measured in terms of the
Search WWH ::
Custom Search