It converts retinal luminance and chroma signals into neural activity patterns it uses in parallel pattern-matching activities. This process may produce numerous hypotheses. Some may involve unwarranted possibilities manufactured by creating absent objects to complete nonexistent patterns.

Since visual input may lack sufficient data, numerous contenders may be equally plausible. Therefore, we are vulnerable to visual misperceptions (illusions) based on incomplete accumulated knowledge and perceptual creativity that suggests incorrect interpretations when the truth is improbable. Worse, in ambiguous situations, the process may never converge and the interpretation may unpredictably alternate between equally likely candidates. Research clearly indicates our upbringing affects our susceptibility to illusions.

Thomas Young’s 1801 Color Experiments

In 1801, Thomas Young noted the well-known fact that mixing any two spectrum colors produced another color that was visually indistinguishable from another authentic spectrum color. This landmark observation is not intuitive and is unique to visual stimulation. As an example, musicians can typically listen to (sense auditory stimuli) musical chords containing numerous individual notes and decompose the chord into individual constituent components (notes). However, this is impossible to do with color blends.

Young also noted he could select three, but not two, widely spaced spectral colors and mix them in varying amounts to produce a wide variety of other colors including many spectral colors, white, and pastels. The three colors did not have to be unique, though for producing the greatest color variations it was imperative that any one of the three colors could not be produced by mixing the other two. He initially used red, yellow, and blue but later changed to red, green, and blue.

From these observations, Young deduced the first version of what is now referred to as the Young-Helmholtz theory. This theory hypothesizes the human eye contains three cone types which respond to red, green, and blue stimuli and all color sensations result from blending the three cone type signals.

This was a significant departure from conventional thinking because it suggested a few receptor types could distinguish many colors. Because each cone type has different sensitivity characteristics, the frequency change required from one spectral color to the next visually different spectral color varies across the visible spectrum as illustrated by Figure 4-2 on the following page. Humans can typically detect 128 fully saturated visual spectrum colors and can detect 50 percent more saturation levels (23 levels) at the visible spectrum ends than at the center (16 levels).

Figure 4-2  Hue discrimination curve

Color Perception Summary

Humans normally can generally distinguish hundreds of thousands, if not millions, of colors derived from various combinations of 500 different luminances (achromatic gray level intensities), 200 hues (individual colors), and 20 shades for a hue (intensities, sometimes called the chroma range). The eye is more sensitive to hue shifts at the extreme ends of the visible spectrum than the interior spectrum colors.

In other words, human detectability of color errors varies by hue, saturation, and intensity. While we might be proud of our visual perception that stems from three sensor types, consider that one humble type of desert salt brine organism has nearly a dozen different color receptor types. Birds, fish, reptiles, and various insects also possess excellent color vision though dogs and cats have none.

White Point

The human vision system is not particularly discriminating about what it considers white light. The reference white point selection process is somewhat elusive but involves general illumination levels. For example, car headlights may seem white at night but yellow during a summer afternoon. Hence, white reference points shift depending on circumstance. And, as it does, human color vision effortlessly and easily adjusts to compensate—referred to as illumination source compensation. Thus, the human visual system’s automatic tendency to strive for consistency produces different color balance sensations in different settings. Among other things, this makes it difficult to select house paint colors. While the paint looks great in the store, it may look hideous on your house.

Flicker

Flicker primarily complicates digital video because CRTs necessarily exhibit some flicker which has historically been linked to electrical power frequency rates. To further complicate matters, Britain and many European countries display 50 television images per second while people in the United States view 60 images per second. In both cases each image remains on its screen portion during the period the next image replaces a previous image on the other screen portion. So, viewers actually simultaneously see two or more interlaced images which interleave line by line. This can complicate digital image processing for television images captured with frame grabbers since the various image portions are time separated by as much as 0.04 second.

Conclusion

Human perception is an enormously complicated subject. The main objective of this chapter is to mention a few points that clearly illustrate perception complexities we regularly take for granted. But understanding how human perception works helps predict or explain the effect of some image processing approaches.