Fusion of Visual Channels - Scientific Visualization: Uncertainty, Multifield, Biomedical, and Scalable Visualization - page 116

see the visual mappings of individual field remains, there is also an additional
constraint, that is, in order to compare field F a and F b , their visual mappings,

V a

and V b have to be visually comparable.

• Observation of Combined Effects —In some applications, different fields can be

combined meaningfully, by using, for example, logical, algebraic and statistical
operators. In such cases, the emphasis has been shifted from the depiction of
individual fields to that of the combined effects.

Note that these three types of tasks are not totally exclusive or performed in
isolation. For example, once an observation of association is made, one may use
combined effects, such as point-wise subtraction, to compare two fields and evaluate
the changes from one to another.

Volume Visualization . Earlier works in volume visualization were focused primar-
ily on logical operators on voxel occupancy in combination of multiple volume
datasets (e.g., [ 8 , 21 ]). At the turn of the new millennium, three groups indepen-
dently developed methods for constructive operators for combining real-domain vol-
ume datasets [ 3 , 6 , 17 ]. Among them, Chen and Tucker [ 6 ] provided an algebraic
framework, CVG, for combining discretely specified volume datasets and contin-
uous functional specifications of scalar fields, and rendering a multi-volume scene
as a CVG expression that is equivalent to a tree or scene graph. The constructive
operators are defined with four separate channels for

, R, G and B, each of which
can be operated on differently from others. The open-source vlib [ 22 ] extended
this approach to allow for independent constructive operators on attribute fields for
the Phong illumination model, reflection and refraction. Chen [ 4 ] later added the
notion of point-based volume objects into the CVG framework to enable scalar fields
defined by point clouds to be visualized with volume datasets and functional scalar
fields.
It is usually difficult to evaluate the characteristics of individual fields through
combined effects in multifield visualization. To support 3D visualization tasks with
strong elements of association and correlation, one critical problem is occlusion.
While translucency is effective for simultaneously depicting non-overlapping fea-
tures in volume visualization, it is necessary to use other visual channels to deal
with features that are from different fields and occupy more or less the same space.
One approach is to use visual representations that exhibit a reasonable amount of
empty space, such as wire frames and coarse non-photorealistic textures. For exam-
ple, one method proposed in [ 20 ] is a class of pen-and-ink textures which can be
used in conjunction with conventional shaded RGB surfaces and

α

RGB amorphous
effects. The density of the pen-and-ink lines and the opacity of the empty area in
the textures can be controlled using a transfer function. Another approach is to use
deformation to force a field to make space for another field, provided that the users
can still comprehend the actual spatial position of the deformed field. For example,
Chen et al. [ 5 ] introduced deformation as a generic operator in the CVG framework
and the detailed displacement as fields that can be applied onto any other scalar fields
in a scene graph to facilitate focus and context views of different scalar fields.

α