Glyph-Based Multi-field Visualization - Scientific Visualization: Uncertainty, Multifield, Biomedical, and Scalable Visualization - page 124

indicate regions and layers of importance (e.g., Rate of strain tensor example empha-
sized the velocity more by using black arrows). This method enables the simultaneous
depiction of 6-9 data attributes, in which the authors apply to a simulated 2D flow
field past a cylinder at different reynolds number. The example shows the visualiza-
tion of velocity, vorticity, rate of strain tensor, turbulent charge and turbulent current.
Visualizing Multiple Fields on the Same Surface by Taylor [ 22 ] provides an
overview of successful and unsuccessful techniques for visualizing multiple scalar
fields on the same surface. The author first hypothesizes that the largest number of
data sets that can be displayed by mapping each field to the following: a unique sur-
face characteristic, applying a different visualization technique to each scalar field
or by using textures/glyphs whose features depend on the data sets. This framework
is limited to visualizing up to four scalar fields. The author then describes two tech-
niques that prove effective for visualizing multiple scalar fields, (1) data-driven spots
(DDS) —using different spots of various intensities and heights to visualize each data
set, and (2) oriented slivers —using sliver like glyphs of different orientations that
are unique to each data set along with various blending.

13.2.2 Spatial Dimensionality: 2.5D

A Scientific Visualization Synthesizer by Crawfis and Allison [ 3 ] introduces a novel
approach for visualizing multiple scientific data sets using texture mapping and raster
operations. The authors present an interactive programming framework that enables
users to overlay different data sets by defining raster functions/operations. Using a
generated synthetic data, the author presents a method for reducing the visual clutter
by mapping color to a height field and using a bump map to represent the vector plots
and contour plots. The final texture is mapped onto a 3D surface.
Peng et al. [ 16 ] describes an automatic vector field clustering algorithm and
presents visualization techniques that incorporate statistical-based multi-variate
glyphs. In summary, the authors clustering algorithm is given by: (1) derive a mesh
resolution value for each vertex, (2) encode vector and mesh resolution values into
R, G, B and

in image space. Clusters naturally form in this space based on pixel
values. (3) The clusters are merged depending on a similarity value derived using
Euclidean distance, mesh resolution, average velocity magnitude and velocity direc-
tion. Several clustering visualizations are given, using

α

| v | -range glyph that depicts the

local minimum and maximum vector, and a -range glyph that shows the variance of

vector field direction along with the average velocity direction and magnitude. Other
visualization options include streamlets that are traced from the cluster centre, and
color coding with mean velocity. The authors demonstrate their clustering results on
a series of synthetic and complex, real-world CFD meshes.

θ