Image Processing Reference
In-Depth Information
green, and blue but rather to perceptions of colors that could be called yellowish-
green, green and blue-violet, respectively. Figure 2.1 illustrates the average sensi-
tivity of the cones to photon wavelengths along with perceptions of colors upon re-
ception of photons with such wavelengths. Note that hues associated with pink/rose
are absent at the bottom of the diagram. This is because there are no photons with
such wavelengths in the nature. Pink is a sensation response of the brain to a mix-
ture of light composed of photons predominantly from short (blue) and long (red)
wavelengths.
The dotted sensitivity curves in the graph are published experimental data [80],
whereas the solid curves are Gaussians fitted by the author.
( ω ω 1 ) 2
2 σ 2
( ω ω 2 ) 2
Long
(570 nm):
(exp(
2 σ 2 )) /C ,
where C =1 . 32, ω 1 = 540, ω 2 = 595, and σ =30.
)+exp(
( ω ω 1 ) 2
2 σ 2 ),
where ω 1 = 544, and σ =36.
Middle (540 nm):
exp(
( ω ω 1 ) 2
2 σ 1
( ω ω 2 ) 2
2 σ 2
)) /C ,
where C E =0 . 75, C =1 . 48, ω 1 = 435,
ω 2 = 460, σ 1 =18and σ 1 =23.
More than half of the cones are L-cones (64%). The remaining cones are predomi-
nantly M-cones (32%), whereas only a tiny fraction are S-cones (4%). It is in fovea,
within appproximately 1 of eccentricity, where humans have the densest concen-
tration of cones. The fovea has no rods, and with increased density towards higher
eccentricities, the cone density decreases while the rod density increases. Even the
cones are unevenly distributed in the central part, with M-cones being the most fre-
quent at the very center, surrounded by a region dominated by L-cones. The S-cones
are mainly found at the periphery, where the rods are also found. The center of the
retina is impoverished in S-cones (and rods). The minimum amounts of photons re-
quired to activate rods, S-cones, M-cones, and L-cones are different, with the rods
demanding the least. Among the cones, our M-type need the least amount of photons
for activation, meaning that more intense blues and reds, compared to green-yellows,
are needed in order to be noticed by humans.
The coarseness of a viewed pattern matters to the photoreceptors too. A retinal
image with a very coarse pattern has small variations of light intensities in a given
area of the retina than does a fine pattern that varies more. A repeating pattern is
also called texture . Coarse textures contain more low spatial-variations than fine tex-
tures. Silhouettes of people viewed through bathroom glass belong to the coarse cat-
egory. A retinal image with “fine” texture is characterized by rapid spatial changes
of the luminosity such as edge and line patterns. This type of patterns is respon-
sible for the rich details and high resolution of the viewed images. We will discuss
the coarseness and fineness with further precision, when discussing the Fourier trans-
form and the spectrum (Chap. 9). Generally, the photoreceptors at high eccentricities,
i.e., basically S-cones and rods, respond to low spatial-variations (spatial frequen-
Short
(440 nm):
(exp(
)+ C E exp(
Search WWH ::




Custom Search