Interactive Visual Exploration and Analysis - Scientific Visualization: Uncertainty, Multifield, Biomedical, and Scalable Visualization - page 155

space to data of interest while presenting the remainder as context for orientation.
However, in a generalized setting, various visual attributes can serve to emphasize
or deemphasize data items, including, for example:

￿ Color (hue, saturation, brightness, or an alternative representation) and opacity: a

typical example would be to present the data subset in focus in color and its context
in gray scale by mapping the degree of interest to saturation [ 8 ]. Alternatively, the
degree of interest can be mapped to opacity, rendering the focus opaque and the
context semi-transparent [ 21 ].

￿ Style: different visualization modalities (isosurfaces, volume rendering, etc.) can

be used to discriminate focus and context. Alternatively, rendering styles, in par-
ticular non-photorealistic/illustrative styles (halos, outlines, cross-hatched/dotted
lines/polygonal primitives) can serve this purpose (for example in a two-level vol-
ume rendering approach [ 11 ]).

￿ Frequency: Only use the full spectrum of spatial frequencies for the data subsets

in focus and render the context band-limited. This approach is called Semantic
Depth of Field [ 18 ] and results in a blurred style for the context, directing the
user's attention to the sharply rendered data subsets in focus.

￿ Space: This approach refers to the traditional notion of F+C visualization, i.e.,

that the visualization space is distorted in order to give more space (or time) to the
visualization of data subsets in focus.

15.3 Levels of IVA

So far, with “Show & Brush”, we have seen the base level of IVA. Based on the
complexity of feature definitions, we distinguish additional, more complex (and
thereby also more powerful) levels of IVA. It is our experience, however, that in many
cases—if not in most cases—the simple Show & Brush technique already provides
sufficient functionality to enable an effective data analysis; the more complex levels
of IVA, as introduced below, are only advanced solutions for more complicated cases
which cannot be served with the base-level IVA.

1. Show & Brush (level 1) : This level captures the analysis as described so far. It
utilizes at least two linked views, usually one physical and one range view. The
interactive selection of features of interest is accomplished by brushing in one
view, leading to a focus plus context visualization in the linked view(s).
2. Relational analysis (level 2): This level supports the combination of brushes
using logical operations and a simple feature definitions language.
3. Complex analysis (level 3): This level integrates computational analysis, e.g.,
derived fields, statistical methods, machine learning [ 28 ], etc., into the interactive
visual approach, thus adding a new dimension of possible procedures. A typical
scenario would be that, prompted by insights gained during visual exploration and
analysis, the user decides to initiate a certain computational analysis procedure,
such as clustering. This procedure results in at least one additional (synthetic)