I would like to thank Dr. Donna L. Gresh for her help in using
the colormap for defining a transfer function. I am doing
"segmentation" of images in which I want to select certain
gray levels. In particular, arteries given slightly higher density
than blood using a injection. The problem is the bones that
block certain angles in the three-dimensional view. With an isosurface,
if we assume continuous functions than between a bone of high density
and tissue of low density there must be a value of intermediate
density, even if it is an artifact. The result is that the isosurface
also appears around the bone. I know how to make slabs to cut the
view but artery winds around and no planar cut does a good job of
improving the view. By selecting a narrow range of the field I
can make the bone have the same data value as the low density tissue.
Well, almost, the bone erodes but a part is still visible. In the
future we may write a filter but I wanted quick solution.
Last night I continued to experiment. In case someone else
wants to do something similar, I will try to describe in words
the steps. In the original net I had a field output and a data
output. The data output was taken from the field using extract.
Now, instead, the path of the field has one module inserted: Replace,
with source name and destination name "data" . With regard to
the "data" path, as in the original the first step is to Extract
the "data" from the field. The next step is to put the data
into a Colormap then combine the original source with the Colormap
output using the module Color. The next module may be superfluous,
I see that I put it in and the network works but perhaps it would
work without it, I am refering to Extract of "colors". In any case,
moving to the next module, I use Compute a.1 + a.2 + a.3 .
Then I used Mark. Donna Gresh suggested instead to use first Mark "colors"
then Computer. Her suggestion seems more logical, I do not know why
the other order works since I use Mark "colors" after I have converted
a three-component field into one component. In any case, either works.
The next step is a branch, the result is sent where before I sent "data"
and also it is sent to the Replace module that I mentioned above.
In this case, I used the color value to control the transfer function.
While I have not tried it, it seem to me to be in interesting possibility
to use each color for separately for different isosurfaces.
Thanks for the help,
Alan Scheinine
