Digital Signal Processing Reference
In-Depth Information
Figure 2.1.
Processing Structures of the eye: although we consider the eye, the
organ of sight, if we are dealing with the projected form, it is a) the retina that
preprocesses the visual information. In b), the plus structures are rods that are
sensitive to light regions and minus structures are rods that are sensitive to the
dark. Through a specialized excitory/inhibition structure within retina, intensity
discontinuity detection can be implemented.
1.
Vision Requires Specialized Hardware
Although we think of the human system has ultimately adaptable, the
visual system has many functionalities that are fixed at the cellular
level. For instance, the retina is an amalgamation of two different
types of cells: rods and cones. The rods absorb a broad spectrum
of light; the cones absorb a color-specific range. Furthermore, the
cones are differentiated by the color absorption of their pigments
to sense trichromaticity. Even within the same types of cells, the
hard-wired connectivity among cells also varies. Different types of
cells have different behaviors toward sustained and transient image
patterns. Biological vision is not a single unified process, but rather
the management of many sources of information toward the visual
objective. When we design our system, a properly managed set of
useful primitives may be closer to our own biological solution.
2.
Vision Has Competing Processes
Differentiation in vision occurs not only at the cellular level, but also
at the process level. Vision itself has two competing processes for
bright and dark conditions. In bright conditions, the cones offer us
color vision. In dark conditions, the rods, which are more sensitive
and have a broader absorption scale, produce most of the visual in-
formation and suppress cone activity. Once again, when we design
our software systems, we should not be surprised to find competing
systems working in parallel.
3.
Vision is Likely Represented in Spatial Coordinates
We have naturally defined our video sequence in terms of spatial
coordinates. However, we may also interpret the data in terms of the
Search WWH ::
Custom Search