Spatial Cues in 3D Visualization - Ambient Intelligence for Scientific Discovery - page 106

represent data or other information), it is useful to draw from established theory
that explains how we naturally and effectively perceive objects in the real world.
How does our perceptual and cognitive systems ascertain the relative depths
and understand the spatial relationships of objects that are viewed in our natu-
ral environment? This issue becomes more complicated when there are multiple
sources of information that may incrementally support (or conflict with) our
cognitive assessments of the relative sizes, positions, and orientations of objects
viewed in space.
Cue Theory. [1-5] maintains that the human visual system infers the distance of
environmental objects based on information relating to the posture of the eyes
and to visual patterns projected onto our retinas. The operative visual depth
cues that serve these purposes are many and varied, and are often placed into
the two categories of primary and secondary cues discussed in the Introduction to
this chapter. The general tenets of cue theory, that these primary and secondary
cues impact our perception and understanding of the relative spatial attributes
of objects viewed in space, are widely accepted. However, the particular pro-
cesses used by the visual system in selecting and/or combining multiple depth
cues to arrive at a singular, stable perception of the depth of objects in space
are widely debated. What happens when there are multiple cues that comple-
ment (or conflict with) each other in providing information about the depth of
objects seen within our visual fields? In these circumstances, there are at least
two countervailing views [3-5], that the perceptual system: (1) selects certain
cues that dominate over other cues; or (2) integrates the multiple available cues,
fostering a perception of depth that is incrementally affected by each individ-
ual cue. Additionally, there are variants of both the selection and integration
approaches.
Two variants of the selection approach include vetoing and fusion models.
In [3], they suggested a vetoing mechanism that is operative when conflicting
information is provided by multiple cues. In these situations, the more dominant
depth cue simply overrides the effect of the weaker cue(s) and the perception
of depth is provided by the stronger cue alone. Furthermore, [3] and [4] discuss
weak and strong fusion mechanisms that combine characteristics of both the se-
lection and integration approaches. In the weak fusion model, depth information
is cognitively processed separately from each cue, and subsequently combined in
a weighted linear manner to produce an overall depth effect. In strong fusion,
there is a non-linear interaction among multiple depth cues. As an example, one
cue may work to disambiguate information, allowing depth information to be
extracted from another cue.
Two variants of the integration approach include additive and multiplica-
tive models. In [5], they provide an comprehensive summary of the additive and
multiplicative depth cue integration approaches. However, the crux of the ad-
ditive model is that depth information provided by multiple individual cues is
aggregated such that additional cues always provide more information about
depth. In contrast, the multiplicative model maintains that there is a synergy
among multiple depth cues which creates either a “greater than” or a “lesser