Biomedical Engineering Reference
In-Depth Information
In this context, different interaction strategies can solve problems normally asso-
ciated with visualization in a more efficient manner. Prominent examples are VR
interaction strategies, e.g. augmented reality, navigation devices or haptic interfaces.
5.4.1 Multiscale Visualization
This section discusses multiscale visualization. It presents the design process, impor-
tant techniques and recent realizations.
Multiscale Design Process
The current multiscale design process consists of choosing the appropriate techniques
to create a multiscale system, depending on the following factors [
18
,
46
]:
Type of data and their features: formats, dimensionality and amount.
Visualization style: e.g. isosurfaces, volume rendering, vector field, tensor field
visualization, … [
47
].
Nature of multiscale: Considerations to be taken into account are based on the rela-
tions between the different types of data, as the order of magnitude or the relation
between space and time scale. If the data does not have time-space continuity, a
smooth transition is required. The presence of these gaps between different scales
is one of the major challenges in the design process.
Style of interaction: navigation, augmented reality, haptic and gesture interaction
(see Sect.
5.5.2
).
Multiscale Techniques
Current multiscale techniques can be categorized by their function [
18
]. The most
relevant of these techniques are:
Out-of-core visualization:
This collection of techniques handles datasets that are
larger than the available memory [
3
]. General external-memory techniques can
be divided in two groups:
batched computations
and
on-line techniques
. The first
group involves data streaming into internal memory. Later the data is processed in
multiple passes. In the second group, also based on batched computations, the data
is pre-processed according to possible queries and results are stored in a specific
structure that facilitates the access.
Level of detail (LoD):
The representation complexity of an object in the scene
depends on its relevance (e.g. position, camera speed or user focus). As datasets
grow in size and complexity, the importance of LoD techniques is increasing [
48
].
Call-outs and lensing:
These two techniques allow simultaneous view of detailed
and global content. Call-outs (Fig.
5.6
a) are enlarged sub-regions that link to a
