Biomimetic Vision Sensors - Engineered Biomimicry - page 21

pole); this assumes a 2 mm pupil and illumina-
tion wavelength of 550 nm [7] . A normal (i.e.,
emmetropic) human eye that measures 20/20 on
a standard eye chart sees with a resolution of
approximately 1 min of arc. Perhaps not coinci-
dentally, the minimum separation of individual
photoreceptors in the fovea [32] and the sampling
theorem requirement of at least two samples per
cycle of spatial frequency [27] together yield a
physiological resolution limit for the eye of just
over 1 min of arc [22] , closely matching the theo-
retical optical resolution limit due to the diffrac-
tion-limited cutoff frequency of a 2 mm pupil in
daylight. Certain types of acuity, such as vernier
acuity, have been shown to exceed this theoretical
resolution, most likely due to higher-level pro-
cessing in the brain's visual cortex [32] .
As mentioned earlier, the camera eye, of
which the human eye is a particularly good
example, is the basis on which nearly all stan-
dard cameras and vision sensors are based.
Some optical system is used to bring an image
to focus on an FPA, which then spatially samples
the image. The FPA typically is rigidly attached
to the sensor frame and cannot move. The focal
length is also typically fixed (except for zoom
lenses). With reference to Figure 1.1 , if some par-
ticular object distance s o is desired, the optical
center (nodal point) of the lens or lens system is
moved axially to ensure that the image distance
s i remains equal to the distance between the
nodal point and the fixed FPA. This simple
camera-eye model can be used to analyze or
design a wide variety of vision sensors. How-
ever, in search of capabilities beyond that of
standard cameras and vision sensors, much of the
recent work in the area of biomimetic vision sen-
sors has turned to models of the compound eye.

seminal work [36] . We then narrow our focus to
the common housefly, M. domestica , and describe
its vision processes in some detail, including the
feature of motion hyperacuity. A review of phys-
ical sensors and navigation systems inspired by
the insect world is then provided. We conclude
with potential applications for these specialized
sensors. This section is based in part on several
previous journal papers on this topic [37-40] .
Compound eyes have been around for some
time. Trilobites featured compound eyes, and
they existed from approximately 500 million to
250 million years ago [41] . Compound eyes are
also quite common in insects. Insect compound
eyes are quite mobile relative to the head. This
allows advanced features such as a wide field of
view and depth perception by employing stereo
vision [41] .
Depending on specific species, a compound
eye includes hundreds to tens of thousands of
individual hexagonal-shaped facet lenses. Each
individual lens is composed of a chitinous-type
material that serves as the cornea for the primary
modular vision unit of the compound eye, called
an ommatidium . The lens ranges in size from 15
to 40 µ m, again depending on species [41] .
The ommatidia form a repeatable pattern
across the curved surface of the compound eye.
The angle between adjacent ommatadia, called
the interommatidial angle , ranges from 1 to 2
degrees, depending on species. As we shall see,
this angle has a large impact on determining the
resolving power of the eye [41] .
After light enters the facet lens, it passes
through and is focused by the crystalline cone.
The cone is made up of a transparent solid mate-
rial and focuses the impinging light on the prox-
imal end of the rhabdom. The rhabdom channels
the light to the photosensitive receptors called
rhabdomeres [41] .

1.3.2 Compound Eye

1.3.2.1 Types of Insect Compound Eyes
Land described three basic configurations of
insect vision: apposition, superposition, and
neural superposition compound eyes [36] .

Many insects and other species have com-
pound eyes. In this section we provide a basic
description of compound eyes followed by three
basic configurations as described by Land in a