This was never really defined in DICOM, prior to GSPS.
When the pipeline for the grayscale image processing was defined
for grayscale presentation state in Sup 33 (see PS 3.4 N.2) a
choice had to be made as to how to describe this, and it was
decided to specify the point transformations (like VOI LUT aka
window center/width) first, and then the spatial transformations
to the resulting P-Values.
The effect of this choice when magnification is implemented
with interpolation is called out in Note 2 of section N.2, but
it was decided that this didn't really matter for all practical
intents and purposes.
So from an implementation perspective it really boils down to
which is more efficient ... interpolation and magnification
of 16 bit data and then having to window more 16 bit values to
8 bits, or windowing 16 bit values to 8 bits first and then having
to interpolate and magnify more 8 bit values.
I could use some good advice.
I am interested in acquiring some low-level development skills that
is relatively hassle-free as far as boards/hardware setup goes since
I am doing this at my own expense.
I would really appreciate some opinions regarding which
OS (Windows variants vs. Linux) would be a better platform to
start with and why? And, embedded vs. device driver?
Someone suggested acquiring embedded Linux (RTLinux) skills but
I'm not sure that these skills are better to acquire over
- Linux device driver skills; or
- Windows variants device driver skills; or
I'm a newbie at low-level development so I hope I'm making sense.
Thank you for your attention.
4. ppp dialer