Biomedical Engineering Reference
In-Depth Information
recognized bottom-up hierarchical theories is that formulated by David Marr in his theory
of vision [59]. By way of contrast, an increasingly popular recognition processing theory
is that of top-down processing. One model, proposed by Moshe Bar [58], describes a
'shortcut' method in which early visual inputs are sent, partially analyzed, from the early
visual cortex to the prefrontal cortex (PFC). Recognition memory can be supported by
both the assessment of the familiarity of an item and by recollection of the context in
which an item was encountered. Some have hypothesized that the PFC disproportionately
contributes to recollection, whereas an alternative view is that the PFC contributes to
both recollection and familiarity. Possible interpretations of the crude visual input is
generated in the PFC and then sent to the IT, subsequently activating relevant object
representations which are then incorporated into the slower, bottom-up process. This
'shortcut' is meant to minimize the amount of object representations required for matching,
thereby facilitating object recognition [58]. Lesion studies have supported this proposal
with findings of slower response times for individuals with PFC lesions, suggesting use
of only the bottom-up processing [60].
A significant aspect of object recognition is that of object constancy, that is to say the
ability to recognize an object across varying viewing conditions. These varying conditions
include object orientation, lighting, and object variability (size, color, and other within-
category differences). For the visual system to achieve object constancy, it must be able
to extract a commonality from the object descriptions from different viewpoints and the
retinal description [61]. Several theories have been generated in order to provide insight as
to how object constancy may be achieved for the purpose of object recognition, including
viewpoint-invariant, viewpoint-dependent
,and
multiple views
theories.
Viewpoint-invariant
theories suggest that object recognition is based on structural infor-
mation, such as individual parts, allowing for recognition to take place regardless of the
object's viewpoint. Accordingly, recognition is possible from any viewpoint, as individual
parts of an object can be rotated to fit any particular view [62]. This form of analytical
recognition requires little memory as only structural parts need to be encoded which, in
turn, can produce multiple object representations through the inter-relations of these parts
and mental rotation [62]. Therefore, storage of multiple object viewpoints is not required
in memory.
Viewpoint-dependent
theories suggest that object recognition is affected by the view-
point at which it is seen, implying that objects seen from novel viewpoints reduce the
accuracy and speed of object identification [63]. This theory of recognition is based on a
more holistic system rather than by parts, suggesting that objects are stored in memory
with multiple viewpoints and angles. This form of recognition requires a lot of memory
as each viewpoint must be stored. Accuracy of recognition also depends on how familiar
the object is from the observed viewpoint [62].
The
multipleview
theory proposes that geometric constraints govern the projection of
points, lines, curves, and surfaces in multiple images. Object recognition lies on a view-
point continuum where each viewpoint is recruited for a different type of recognition.
At one extreme of this continuum, viewpoint-dependent mechanisms are used for within-
category discriminations whereas, at the other extreme, viewpoint-invariant mechanisms
are used for the categorization of objects [63].
Some models for object recognition have been suggested such as the 3D model rep-
resentation, proposed by Marr and Nishihara [64], who state that object recognition is
