Human Perception

We have covered the principles of light, color creation using additive and subtractive methods, and how the eye captures images. This next section addresses the important topic of how images are constructed and what factors affect color perception. We start with an explanation of important terms, discuss cones and color vision in more detail, and then move on to other elements of human perception that are important for image processing. Finally we review Thomas Young’s revolutionary color experiments. A complete treatment of perception is not possible since the science is not completely understood. Therefore, this section is a superficial overview.

Definition of Important Terms

Intensity, or more precisely radiance, is the total EM power that an energy source transmits onto a surface area, measured in watt units. Camera light meters that measure intensity to calculate proper exposures for a given film is a practical light intensity measurement example. Radiometers are a more general measurement instrument.

When the human visual system processes arriving EM energy, it expresses perceptual light intensity differences as luminance differences. Luminance is an objective metric calculated using the light frequency distribution of a source light, called an illuminant, and the human eye’s sensitivity to various wavelengths, called the spectral sensitivity function. Luminance is expressed in lumen units, and humans can determine side-by-side luminance differences with an accuracy of about 1 percent. The luminance difference between areas of highest and lowest luminance is called the contrast.

Intensity and luminance have dramatically different meanings as far as the human visual system is concerned. By definition, one lumen of blue seems to have the same strength as one lumen of yellow-green light. However, 1 watt of 555 nm yellow-green radiant power provides 680 lumens and 1 watt of blue light is approximately 70 lumens. This means that it provides only about 3 percent of the sensation strength that a yellow-green light produces. This color sensation strength difference is due to the long and medium wavelength (red and green) cones collectively being much more sensitive than the short wavelength (blue) cones.

Brightness is a subjective visual response to EM energy and depends on the current adaptation state of the eye. For example, a lighted match has a constant radiation level. While it may appear “bright” in dark rooms, it is nearly invisible in “bright” sunlight. Hence, the term brightness is considered subjective and possesses little technical value.

Spectral Sensitivity

In 1801, Thomas Young began experimenting with color, and based on his observations the Young-Helmholtz Theory became a landmark discovery in the field of color science. The Young-Helmholtz Theory determined that the human retina contains three types of cones. Each cone is sensitive to a wavelength range known as its spectral response function. Although each cone has a fairly wide spectral response function, Young discovered that each of the three cone types is particularly sensitive to light at certain wavelengths. This was a significant departure from conventional thinking because it suggested that a few receptor types could distinguish many colors while contemporary conventional wisdom suggested different sensors sensed each color.

One type of cone, the S-cone, is most sensitive to light at 440 nm (blue). Another, the M-cone, favors light at 545 nm (yellow-green), and a third, the L-cone, is most sensitive to light at 580 nm (yellow-red). These spectral response functions are shown in Figure 2-2. Notice that the blue spectral response function is isolated from the red and green which themselves significantly overlap. This significant overlap causes color reproduction problems because it is impossible to stimulate either of the L-cones or M-cones without stimulating the other—causing ambiguous sensing. Collectively, the three cone types experience a peak sensitivity at around 550 nm (green). A plot of the collective sensitivity is called the CIE luminous efficiency function.

Figure 2-2  Spectral response functions

Individuals (mostly males) who are color-blind have fewer than three cone types (they are not tritanopes). A color-blind person lacks the spectral information to distinguish between certain pairs of colors. Restated, color-blind people have more metamers (apparent color matches) than normal people. Because even normal people cannot distinguish between metameric color mixtures, you could say that everyone exhibits some color-blindness. In fact, S-cones appear to be absent in the fovea center. So, tiny blue objects or text characters on monitors are not perceived correctly or only with difficulty.

Looking at Figure 2-2 we can see why the luminance of a blue light is lower than the luminance of an equal intensity yellow-green light—the spectral response functions show that the green and red cones are collectively more sensitive than the blue (short wavelength) cones. Television and image compression schemes exploit this relative insensitivity by reducing the blue spectrum fidelity.