Conclusions
5. Conclusions
The ability to extract abstract shape properties is a fundamental requirement
for visual perception. In order to extract such properties, the visual
processing is divided into two stages. In the first stage, a "base
representation" is constructed using a bottom-up approach and relying on
the visual input alone. Top-down prcocessing is then applied to the
base representation through the execution of visual routines, in order
to determine definitions of objects and spatial relations amongst them.
Visual routines themselves are composed of a fixed set of elemental operations
which can be combined to perform specific computations and provide specific
goals. Incremental representations store the results after applying
visual routines to portions of the base representation and can be used
by future routines thereby allowing them to be more efficient. Several
basic elemental operations were introduced, including shifting of processing
focus, indexing, boundary tracing and activation and marking however all
operations employed by the visual system are currently unknown!
Finally, perception of spatial relations appears to be very fast and
simple for humans. We are constantly determing such relations without
without any conscious awareness and "without any thought". However,
taking a moment to think about how a simple relation such as determining
whether some polygon is closed may make one aware that there are actually
some complex processes occuring, all at a fraction of a second! Furthermore,
consider that even with todays best and most powerful computing technology,
we are not even close to imitating the human visual system. Our visual
system is clearly astonishing!