13.9. Eye spectral response
There are two basic types of retinal photo-receptors: cones, responding to bright-light conditions, and rods, responding to low-intensity light. Depending on their spectral sensitivity, the former belong to either L (long-wavelengths sensitive), M (mid-wavelengths sensitive) or S (short-wavelengths sensitive) cones. By combining their separate inputs, the brain creates colors. The cones are concentrated in the center of the retina (fovea), where they are as small as ~2μ in diameter. That puts the angular size of the smallest individual foveal cones at ~1/3 arc minute. Cones become larger - up to about four times - toward outer areas of the retina.
Therefore, eye spectral response is directly related, and influenced by illuminance levels (light intensity) to which it is exposed. Illuminance level determines the level of activity of cones and rods, and with it main characteristics of human vision (FIG. 248).
Note that there is no general agreement on the illuminance level that effectively inactivates cones or rods. Estimates of the mesopic range vary by tenfold, or more, on both, low (0.001 vs. 0.01 cd/m2) and high end (0.6 vs. 10 cd/m2 or even higher). Two recent models are fairly close in their estimate of the low end (~0.01 cd/m2, or somewhat lower, depending on the type of light), but not at the high end: one puts it at about 0.6cd/m2 (X-model, Rea et al. 2004, based on results of He et al. 1997/8), and the other at 10 cd/m2 (Mesopic Optimisation of Visual Efficiency, or MOVE model of the European research consortium, Eloholma and Halonen 2006). Several others, mainly compromising models also have been proposed.
It is helpful to clarify some basic terminology related to this subject, since it can be confusing. While the physical light intensity is measured with radiometric units, its human perception is subjected to eye's selective reaction to it, which is measured in photometric units. The basic radiometric unit is watt (W), and photometric candela (cd), with the latter defined as the luminous intensity produced by a monochromatic light source emitting radiant (photon) flux of 1/683 watt at 555nm into the solid angle of 1 steradian (sr). The original definition of one candela (which was called candle, or candlepower) was the light intensity emitted by a plumber's candle made to specified standards.
Closely related to candela is the luminous flux unit, lumen (lm), defined as a flux (i.e. photon density) corresponding to 1cd of luminous intensity (thus, 1cd=1lm/sr, and a point radiating freely at 1cd luminous intensity produces a 4π lm luminous flux). And the illuminance unit lux is defined as 1lux=1lm/m2.
summarizes these units and related terms.
of cones and rods varies with the wavelength, within so called
spectrum, extending from ~370nm to ~730nm. Energy level
corresponding to the wavelength of light wave - inversely proportional
to the wavelength, and in proportion to the frequency - stimulates
eye photoreceptors, which send received stimulus to the brain. Specific
combinations of stimuli from the three different cone receptor types produce an input from which the
brain creates perception of color.
Photon of light has the energy E=hν, h being the Plank's constant, h=6.6256x10-34 J (Joul), with 1J=6.2418x1018 eV (electron-volts), and ν the frequency, a number of waves per unit of time usually expressed in Hertz, as number of complete oscillations per second (so ν=c/λ Hz, c being the speed of light per second and λ the wavelength, in the same unit).
In general, eye sensitivity to light increases exponentially with the decrease in light intensity, with the wavelength of peak sensitivity shifting from ~550nm in day-light conditions, to ~510nm in total darkness. As illuminance decreases, cone function transforms toward more effective light collection (elevated pigments level, suppression of lateral inhibition, and convergence of individual outputs) at a price of lower acuity. Bellow certain illumination level, cone function enters dormancy, but cones are still ready to respond to a sufficiently intense light.
On the other hand, decreasing illumination level stimulates accumulation of rhodopsin (pigment) in rods, which was washed out at higher illumination levels. It activates rods, enabling the eye to respond to light stimuli of much lower intensity. As mentioned, this gradual shift in eye sensitivity mode goes through three main stages: (1) photopic, in bright light conditions, (2) mesopic, in low light conditions, and (3) scotopic in near total darkness (FIG. 249).
It should be noted that these sensitivity figures assume the same level of light intensity. As already mentioned, photon energy is inversely proportional to the wavelength, but what ultimately determines magnitude of response over the visible spectrum - assuming no selective absorption in the imaging system - is the spectral intensity distribution of the emission. For the sunlight, it peaks at about 500nm wavelength, with about 10% decrease at the violet end and about 20% decrease at the red end of the visible spectrum. Stellar peak emissions can be shifted significantly to either side and, to a smaller extent, planetary emissions (for instance, typical reflectance from Mars' reddish iron oxide areas is about 2.5 times higher at the red end of the visible spectrum than in the green).
Mesopic eye function is considerably more complex than either photopic or scotopic, due to the simultaneous input of cones and rods, both only partly activated. As a result, there is no agreement (within U.S. or internationally) about its exact modeling. It is most commonly presented as a simple sum of the cones and rods functions, as Vmes=xVpho+(1-x)Vsco, respectively, with x varying with illumination level from 0 at the low, to 1 at the high of mesopic range. As illumination continues to decrease bellow photopic level, this theoretical curve (FIG. 249 top right) gradually shifts from the photopic to scotopic curve, maintaining a single peak, and similar overall shape.
However, more recent empirical evidence suggests that the actual mesopic sensitivity follows more complex patterns, with a well developed mesopic curve having two main peaks, one converging toward peak rods sensitivity, the other toward yellow range of the spectrum (FIG. 249 bottom right). Such outcome is logically plausible, since both types of photoreceptors are active, widening eye sensitivity curve. The shift of the cone peak from green to yellow is caused by their relative sensitivity under reduced illumination increasing more in the red and blue than in green/yellow (FIG. 249 bottom left; keep in mind that the plots to the right show normalized sensitivities - rods sensitivity is generally much higher). Simply summing up the adjusted photopic and scotopic function also neglects the fact that the cone sensitivity in the mesopic light conditions also increases relative to that in the full photopic mode. Similarly to the rod function, but to a smaller extent, this increase in sensitivity comes at a price of lowered acuity.
Both mesopic plots in FIG. 249 (right) approximate foveal retinal sensitivity as combined sensitivity of cones and rods. However, actual sensitivity varies from predominantly cone sensitivity in the inner fovea (particularly foveola), to predominantly rods sensitivity toward the outskirts of fovea and further off. In other words, within approximately inner half of the fovea mesopic sensitivity is approximated with the right mesopic wing on the bottom plot, which is higher toward both, red and blue wavelength, than cone sensitivity in the photopic mode. Toward outer portions of foveola, and beyond, where rods become dominant, relative sensitivity increases for medium to short (green-to-violet) wavelengths, and decreases for longer (red) wavelengths, nearly vanishing for the deep red; mesopic sensitivity for this portion of the retina is approximated mainly by the left mesopic wing on the bottom right plot on FIG. 249.
Obviously, sensitivity properties of the retina vary greatly with its particular area. Its small off-center portion, foveola, is nearly exclusively covered by a dense cone structure; cones remain dominant up to about half the radius of fovea, an area surrounding foveola, 4-5 times larger in diameter. Outer half of the fovea has a mixed cones/rods structure.
Rods begin to dominate at the outskirts of foveola, nearly exclusively populating the outer area of the retina. They become predominant at less then 1mm away from the retinal center, and peak at about 15° from the it. Similarly to the cones, their size varies from the smallest - about 2.5 microns - in the area of highest concentration, to roughly double that size far out. Rods are more numerous than cones in roughly 100:1 ratio in the retinal area out of the <2mm of central fovea (FIG. 250). In the entire retina, rods outnumber the cones approx. 20:1 (120 millions vs. 6 millions). Despite similar average size, rods have much poorer resolution than cones. This is mainly a consequence of so called "convergence": neural output from several rods converge into a single neural brain connection, as opposed to the cones, which have individual neural outputs. Neural convergence of the rods improves sensitivity, sacrificing the resolution.
FIGURE 250: Distribution of photoreceptors on the retina of the human eye. The total field of view, approximately 140° is constructed through the principal point (P) of the eye. The cones (yellow line), sensitive to bright light, have highest concentration in the small area (foveola) of ~1/3 mm (1°) in diameter, shifted nearly 12° from the optical center. The three cone types, L (most sensitive to longer visual wavelengths), M (mid wavelengths) and S (short wavelengths), have different color pigments, providing input from which the brain creates sensation of colors. The S cones are by far the least numerous; also, they are least sensitive, with their main function being supplying the brain with needed input to create color blue. Since the S cones are entirely absent from the central ~0.1mm of fovea, this spot is blue blind (it has so called tritanopic vision, in which blue wavelengths are seen as green). The rods (violet line), active in low light conditions, are absent from the central ~1/3 mm of fovea (foveola), but their concentration quickly rises toward the edges of macula, and farther out, to reach the maximum some 18° off the foveal center. While rods, similarly to cones, also differ in size depending on their position on the retina (generally being larger in the outer areas), they only have a single pigment, rhodopsin. It is ultra-sensitive to light in the green/blue wavelength range but, being a single pigment available in low-light conditions, doesn't allow to the brain to create sensation of color. As the surrounding light intensifies, rhodopsin level diminishes, until rods gradually deactivate and cones, also gradually, take over. And vice versa, as the light intensity decreases toward low-light conditions, it falls below the threshold of the cones, while the rods get activated, and become dominant visual receptors. For illustration, size of Jupiter's disc on retina, when magnified 200 times, is shown next to the areas dominated by the cones (smaller than Jupiter's disc), mixed, and rods dominated.
The above comparison of the common size of Jupiter on the retina with the size of its respective cones- and rods-dominated areas suggests that centering an object that requires high resolution in the field of view significantly improves its definition. On the other hand, averted vision will be more helpful with objects where detection is more important than resolution - as long as they radiate mainly in the green-to-violet portion of the visible spectrum. The highest rods acuity is at about 4° from foveola, near to half-radius of the macula. So, for detection of very faint objects, it should work best with the eye directed some 4° off the field center (about 1/5 and 1/10 of the field radius in 40° AFOV and 80° AFOV eyepiece, respectively), with the object centered in the field.