Color constancy


Color constancy is an example of subjective constancy and a feature of the human color perception system which ensures that the perceived color of objects remains relatively constant under varying illumination conditions. A green apple for instance looks green to us at midday, when the main illumination is white sunlight, and also at sunset, when the main illumination is red. This helps us identify objects.

Color vision

is how we perceive the objective color, which people, animals and machines are able to distinguish objects based on the different wavelengths of light reflected, transmitted, or emitted by the object. In humans, light is detected by the eye using two types of photoreceptors, cones and rods, which send signals to the visual cortex, which in turn processes those colors into a subjective perception. Color constancy is a process that allows the brain to recognize a familiar object as being a consistent color regardless of the amount or wavelengths of light reflecting from it at a given moment.

Object illuminance

The phenomenon of color constancy occurs when the source of illumination is not directly known. It is for this reason that color constancy takes a greater effect on days with sun and clear sky as opposed to days that are overcast. Even when the sun is visible, color constancy may affect color perception. This is due to an ignorance of all possible sources of illumination. Although an object may reflect multiple sources of light into the eye, color constancy causes objective identities to remain constant.
D. H. Foster states, "in the natural environment, the source itself may not be well defined in that the illumination at a particular point in a scene is usually a complex mixture of direct and indirect distributed over a range of incident angles, in turn modified by local occlusion and mutual reflection, all of which may vary with time and position." The wide spectrum of possible illuminances in the natural environment and the limited ability of the human eye to perceive color means that color constancy plays a functional role in daily perception. Color constancy allows for humans to interact with the world in a consistent or veridical manner and it allows for one to more effectively make judgements on the time of day.

Physiological basis

The physiological basis for color constancy is thought to involve specialized neurons in the primary visual cortex that compute local ratios of cone activity, which is the same calculation that Land's retinex algorithm uses to achieve color constancy. These specialized cells are called double-opponent cells because they compute both color opponency and spatial opponency. Double-opponent cells were first described by Nigel Daw in the goldfish retina. There was considerable debate about the existence of these cells in the primate visual system; their existence was eventually proven using reverse-correlation receptive field mapping and special stimuli that selectively activate single cone classes at a time, so-called "cone-isolating" stimuli.
Color constancy works only if the incident illumination contains a range of wavelengths. The different cone cells of the eye register different but overlapping ranges of wavelengths of the light reflected by every object in the scene. From this information, the visual system attempts to determine the approximate composition of the illuminating light. This illumination is then discounted in order to obtain the object's "true color" or reflectance: the wavelengths of light the object reflects. This reflectance then largely determines the perceived color.

Neural mechanism

There are two possible mechanisms for color constancy. The first mechanism is unconscious inference. The second view holds this phenomenon to be caused by sensory adaptation. Research suggests color constancy to be related changes in retinal cells as well as cortical areas related to vision. This phenomenon is most likely attributed to changes in various levels of the visual system.

Cone adaptation

Cones, specialized cells within the retina, will adjust relative to light levels within the local environment. This occurs at the level of individual neurons. However, this adaptation is incomplete. Chromatic adaptation is also regulated by processes within the brain. Research in monkeys suggest that changes in chromatic sensitivity is correlated to activity in parvocellular lateral geniculate neurons. Color constancy may be both attributed to localized changes in individual retinal cells or to higher level neural processes within the brain.

Metamerism

Metamerism, the perceiving of colors within two separate scenes, can help to inform research regarding color constancy. Research suggests that when competing chromatic stimuli are presented, spatial comparisons must be completed early in the visual system. For example, when subjects are presented stimuli in a dichoptic fashion, an array of colors and a void color, such as grey, and are told to focus on a specific color of the array, the void color appears different than when perceived in a binocular fashion. This means that color judgements, as they relate to spatial comparisons, must be completed at or prior to the V1 monocular neurons. If spatial comparisons occur later in the visual system such as in cortical area V4, the brain would be able to perceive both the color and void color as though they were seen in a binocular fashion.

Retinex theory

The "Land effect" refers to the capacity to see full color images solely by looking at a photo with red and gray wavelengths. The effect was discovered by Edwin H. Land, who was attempting to reconstruct James Clerk Maxwell's early experiments in full-colored images. Land realized that, even when there were no green or blue wavelengths present in an image, the visual system would still perceive them as green or blue by discounting the red illumination. Land described this effect in a 1959 article in Scientific American. In 1977, Land wrote another Scientific American article that formulated his "retinex theory" to explain the Land effect. The word "retinex" is a portmanteau formed from "retina" and "cortex", suggesting that both the eye and the brain are involved in the processing. Land, with John McCann, also developed a computer program designed to imitate the retinex processes taking place in human physiology.
The effect can be experimentally demonstrated as follows. A display called a "Mondrian" consisting of numerous colored patches is shown to a person. The display is illuminated by three white lights, one projected through a red filter, one projected through a green filter, and one projected through a blue filter. The person is asked to adjust the intensity of the lights so that a particular patch in the display appears white. The experimenter then measures the intensities of red, green, and blue light reflected from this white-appearing patch. Then the experimenter asks the person to identify the color of a neighboring patch, which, for example, appears green. Then the experimenter adjusts the lights so that the intensities of red, blue, and green light reflected from the green patch are the same as were originally measured from the white patch. The person shows color constancy in that the green patch continues to appear green, the white patch continues to appear white, and all the remaining patches continue to have their original colors.
Color constancy is a desirable feature of computer vision, and many algorithms have been developed for this purpose. These include several retinex algorithms. These algorithms receive as input the red/green/blue values of each pixel of the image and attempt to estimate the reflectances of each point. One such algorithm operates as follows: the maximal red value rmax of all pixels is determined, and also the maximal green value gmax and the maximal blue value bmax. Assuming that the scene contains objects which reflect all red light, and objects which reflect all green light and still others which reflect all blue light, one can then deduce that the illuminating light source is described by. For each pixel with values its reflectance is estimated as. The original retinex algorithm proposed by Land and McCann uses a localized version of this principle.
Although retinex models are still widely used in computer vision, actual human color perception has been shown to be more complex.

Retinex

Here "Reprinted in McCann" refers to McCann, M., ed. 1993. Edwin H. Land's Essays. Springfield, Va.: Society for Imaging Science and Technology.