Image Processing Reference
In-Depth Information
and much experience is required to use it effectively. We will present examples
produced by volume rendering in Fig. 7.14.
7.8.2 Selection of parameters
Let us present several examples of ideas concerning parameters in Eq. 7.12
[Drebin88]. Parameters are roughly divided into two groups
{
b
k
}
and
{
α
k
}
.
(a) Selection of
{b
k
}
Values of parameters
determine the basic color tone in rendering and are
called
color codes
or
color values
. They are usually fixed by giving relative
ratios of three components R, G, and B in color codes. We may select differ-
ent arbitrary values so that they correspond to different objects or different
physical phenomena. Examples are shown below.
(i) Color values are changed corresponding to the range of density values in
an original image or a histogram of density values of an original image.
In a medical CT image, for example, the range of CT values varies cor-
responding to each organ. Thus we will be able to distinguish organs by
color in a rendered image. If some abnormal shadows are distinguishable
by CT values, their existence may be noticed easily by changing colors of
voxels having CT values to which attention should be paid in diagnosing
a rendered image.
(ii) Colors may be changed by a position in 3D space. This makes it easy to
observe the state of the density value distribution in a region of interest
and its vicinity.
{
b
k
}
Both methods are used to mark a place to be noticed. The name
color code
comes from this type of usage. In other words, this is regarded as performing
a kind of clustering of voxels beforehand and then utilizing resultant clusters
of voxels for the following processing.
(b) Selection of
{α
k
}
Usually we set this group of parameters so that the special distribution of
density values in a 3D image may be observed clearly as well as possible.
Basically a value of
α
k
is selected to be nearer to
1
at a point satisfying a
condition that we want to see on a rendered 2D image. In other cases we select
values of
empirically with observing resulting 2D images. As an example,
if any one of
α
k
+1
,α
k
+2
,
{
α
k
}
···
,α
k
,α
u
1
, for example, then
1
−
α
u
0
.
Therefore,
β
k
0
. This means
that the value of
b
k
does not contribute to
I
out
(
K
) significantly. That is, the
value of
b
k
does not reach a displayed image plane. This corresponds to the
case that opaque material exists at a point P
u
on the way from a point P
k
to
a display image plane.
(
1
−
α
k
)(
1
−
α
k−
1
)
···
(
1
−
α
k
+1
)(
1
−
α
k
)
Search WWH ::
Custom Search