Understanding Light and Color Vision
Part 2 in a 5-part series on color measurement fundamentals
Color measurement is not intuitive. An object’s physical attributes, such as length and weight, can be easily measured by a calibrated instrument. We might not all agree on whether objects feel light or heavy, but we can all agree on the physics underlying the units of measurement, i.e. pounds or kilograms. The measurement of color, “colorimetry,” is not rooted in physics, but in psychophysics, the branch of psychology concerned with quantitative relations between physical stimuli and their psychological effects. In other words, if you want to quantify the difference between two colors, you show those two colors to as many people as you can then ask them if they are different. You cannot measure color with a ruler because we all perceive color differently. For this reason, color measurement is inherently complex.
Our experience with color—our perception of it, emotional response to it, and preferences for it—all occur in the brain. Visual color matching is still a common practice in the printing industry and contributes to poor color reproduction accuracy and inefficient production. Operators each perceive color differently, no matter how skilled they are in adjusting printing presses and blending ink to achieve accurate color. Factors, such as age, ambient illumination and physical wellbeing, all affect our perception of color.
In general, color is a function of three factors: light source, object, and sensor/detector. Light from a source illuminates an object and the object attenuates the light (reflects it, absorbs it, transmits it, or scatters it). Finally, the attenuated light is collected and measured by a detector. The detector can be anything from the human visual system, to a camera, or a photodiode. This system is illustrated in Figure 1.
Manufacturing industries (photography, printing, paint, etc.) needed a system for efficiently measuring and verifying color. The solution was the CIE (the “International Commission on Illumination”) system of colorimetry, developed over many decades beginning in the 1920s. The goal of colorimetry is to model the perception of an average human observer under a variety of “standard” conditions. These conditions include different types of illumination, such as tungsten lamps, daylight, and fluorescent lamps, and viewing area differences, such as viewing a magazine and viewing products in a grocery aisle. Changing any one of these factors can impact the appearance of an object. The models incorporated into CIE colorimetry were developed from experiments in which participants were asked to view and make judgements about colors under controlled conditions. The resulting colorimetry system allows for the modeling of color perception using three parameters:
- Spectral reflectance (discussed in the previous article in the series)
- Standard observers: mathematical descriptions of human responses to colored light
- Illuminants: mathematical descriptions of “average” lighting conditions
The illustration in Figure 1 is reimagined with the CIE components— illuminant, spectral reflectance, and standard observer, in Figure 2. These three components are run through a color perception model called the CIE L*a*b*. The CIE L*a*b* system is a model of how color is perceived in the brain, a quantified psychological response to the light reflecting off an object and detected by the eye. The perceived difference between the color of two objects is further modeled using color difference.
The following sections discuss in more detail how color is detected and processed in the human visual system, the standard observer, sources and illuminants, the CIE L*a*b* system, and color difference metrics.
The Human Visual System
The perception of color begins in the eye. Light enters the eye through the pupil and is focused by the lens onto an array of cells covering the back of the eye called the retina. The cells in the retina produce an electrical signal when they detect a photon, a quantity of light, and send that signal to the brain through the optic nerve. The retina contains two different types of light-sensitive cells: rods and cones (so named because of their shape; see Figure 3). Rods are primarily used for the detection of light and dark and are much more sensitive to light than cones. The vision we have at night, when light levels are low and color cannot be perceived, is primarily the result of rod vision.
Cone cells are what allow us to perceive color. There are three types of cone cells: long-wavelength-sensitive, medium-wavelength-sensitive, and short-wavelength-sensitive. We’ll refer to them as red-sensitive, green-sensitive, and blue-sensitive, for simplicity (although their sensitives are not exactly aligned with those colors). Cone cells also encompass most of our daytime vision. Rods are too sensitive to most light levels occurring during the day and remain inactive until light levels decrease.
Cone cells can also be thought of as our “high-resolution” vision. Each individual cell views an area smaller than that of the rods. Most cone cells are in a very small area of the retina called the fovea, which encompasses approximately 2-degrees of visual area (basically, the words you are reading right now). An illustration of the fovea and an image of the cone matrix in the fovea (falsely colored to show the different cones) is shown in Figure 4.
There are cones outside the fovea, but they are fewer in number. The consequence of having most of the cones packed into the fovea is that our high-resolution vision, what vision scientists call “high acuity”, is in the center of our visual field. This is the reason we must point our eyes at the words we want to read. Everything in our periphery is blurry, which is okay for detecting motion (e.g. predators), but not for reading. In addition, our color perception is strongest in the center of our field of vision since the cones are used to perceive color. Our perception of color is slightly different for smaller areas on which we focus our vision, like the color of a cereal box, than for larger global areas, like the whole cereal aisle in the grocery store. The concept of viewing area will come up again in the discussion of the two-degree and ten-degree observers.
The eyes are only the entry point to the visual system. After images are detected by retinal cells, the signal passes to the brain for processing, much like the signal from a CCD or CMOS sensor must be sent to a computer for processing. Each half of the visual field (the stuff we look at) is processed by the opposite side of the brain. Everything on our left is processed by the right hemisphere of the brain, and everything on our right is processed by the left hemisphere of the brain. A signal from the eye passes through the optic nerve and into a part of the brain called the “optic chiasm,” where it is divided into left and right visual fields and passed to the right and left side of the brain, as illustrated in Figure 5. A series of processing steps called “optic radiation” takes the signal from the front of the brain to the back of the brain for processing in an area called the occipital lobe, also known as the “visual cortex” (don’t worry, you won’t be quizzed).
The visual cortex is essentially where low-level image processing occurs: edge detection, contrast, color perception, etc. From there it is passed to other parts of the brain for more complicated processing, such as motion detection, facial recognition, and reading.
The models of color vision used to develop colorimetry focus on the color processing that occurs in the visual cortex. If you already have some familiarity with the CIE L*a*b* system, then the ideas discussed here should seem familiar. Once in the brain, the signal from the red, green, and blue cones is transformed into a new color space with channels containing red-green, blue-yellow, and light-dark information, as shown in Figure 6. This is referred to as color opponency, or the opponent-theory of color vision.
The red-green and blue-yellow channels contain no lightness information and cannot comprise an image alone. Vice versa, the light-dark channel contains no color information. Taken together, a full color image is constructed. The breakdown of the perceived object into red-green, yellow-blue, and light-dark channels is illustrated in Figure 7.
The Standard Observer
Standard observers are mathematical functions that describe how the eye detects different colors. Nowadays, there are techniques and equipment used to measure the sensitivity of each cone to different wavelengths of light. Unfortunately, that technology did not exist in the early 20th century, so scientists created psychological experiments (people viewing colors) to devise functions that describe how our eyes detect color. The average results from the experimental participants were calculated to derive the standard observer functions.
Several experiments were conducted over the years to calculate different types of standard observer functions. As previously mentioned, the fovea encompasses a two-degree visual field, which is roughly the area of your thumbnail held at arm’s length (see Figure 8, top). The first standard observer was created in 1931 from experiments in which participants viewed colors that covered a two-degree field of view. Another standard observer was created in 1964 from experiments in which observers viewed colors that covered a ten-degree field of view, roughly the size of your fist, held at arm’s length (see Figure 8, bottom), but with the central two-degree area obscured. Color values calculated using the 1964 ten-degree standard observer are slightly different from those calculated using the two-degree observer due to differences in color signal in the fovea, and area surrounding the fovea.
Sources and Illuminants
As one of the three building blocks in colorimetry, illumination has a significant impact on color perception. Like humans, every light source is unique. While different illumination technologies are great for designers, and the varying shades of daylight make for a dynamic world experience, the lack of standardization means it is very difficult for two people to discuss color unless they are standing in the same room. Therefore, like the standard observer, the CIE created a set of mathematical functions describing average sources, called illuminants. While sources are real objects, illuminants are just numbers.
The color of both sources and illuminants are described using a quantity called “spectral power distribution,” the relative amount of light emitted at each wavelength. While spectral reflectance is normalized to a perfect white, spectral power is usually normalized at 560 nm, roughly the center of the spectrum. Illuminants, the spectral power of average light sources, all have values of 1.0 at 560 nm. The spectral power distributions for a set of “daylight” illuminants are shown in Figure 9. The daylight series of illuminants describe the spectral power of various typical daylight conditions measured off a white object. They are named using their color temperature, a quantity relating to the color of a blackbody (think of the electric range on a stove) as it is heated to different temperatures. Low color temperatures, around 2000K (kelvin), have a reddish glow. As the temperature increases, the color transitions from red, to orange, to white, to richer blues. While daylight is not the same as a blackbody, color temperature is determined by finding which blackbody temperature results in a color most like a specific shade of daylight. Daylight at 6500K is bluer than a daylight at 5000K. The illuminants D65 and D50 refer to 6500K and 5000K daylight, respectively.
In addition to daylight illuminants, the CIE also standardized illuminants for tungsten lamps, Illuminant A, and various types of fluorescent lights, the F-series illuminants. A specific F-series illuminant, F8, represents fluorescent lamps that simulate D50 illumination.
Illuminant D50 has largely been agreed upon as the standard viewing illuminant in the printing industry. True D50 illumination cannot be created using real sources. However, sources such as fluorescent lamps can be manufactured, that have perceived colors close to D50, albeit with different spectral power distributions. Illuminant F8, the spectral power of which is shown in Figure 11 along with Illuminant D50 and a 5000K blackbody, can be used as a reference for manufacturing fluorescent lamps that simulate D50 illumination.
Color Measurement Condition
This is an appropriate point for an aside about measurement condition, which, although it does not directly affect the SpectraLab color measurement system, is important to consider. Spectral reflectance is invariant with respect to illumination (except when fluorescence is involved), and therefore its effect is neutralized by the normalization of target signal to white tile signal. That means the reflectance of an object will be the same whether an instrument uses a LED, Xenon, or Tungsten lamp, all of which have different spectral power distributions.
However, reflectance is not illumination invariant for fluorescent samples. The presence of OBAs in many papers and some white inks is of primary concern. . Fluorescent materials absorb energy in short wavelengths and reemit it at longer wavelengths (high energy to low energy). OBAs, for example, absorb UV energy (high energy) and fluoresce blue light (lower energy). The response of the fluorescent dyes or pigments to energy is called “excitation” and the resulting fluorescence is called “emission.” Illuminating a fluorescent object with more energy causes greater excitation, which results in greater fluorescence. Returning to the OBA example, illuminating a paper sample with more UV energy will result in more blue fluorescence. By extension, instruments with different types of lamps will produce different reflectance measurements, which is a product of both the reflected light and fluorescent emission.
The large variety of color measurement instruments with different types of illumination meant that there was no method to verify that the color measurement of fluorescent objects would be the same when measured at different locations. The problem was solved by the creation of four standard color measurement conditions, classifying different types of illumination: M0, M1, M2, and M3. A description and primary use of the four conditions is provided in Table 1. M0 is a general-purpose condition describing color measurement instruments with uncontrolled illumination. Instruments measuring in M0 should not be expected to produce similar measurement of fluorescent objects. M1 specifies that the illumination spectral power and UV content match those of illuminant D50. This ensures measurements of fluorescent paper and fluorescent colors will be similar for instruments in the M1 condition. M2 specifies no UV be included in the illumination, usually by the placement of a UV-cut filter in front of the lamp. However, white-light LEDs do not produce any UV either, so they can also be classified as M2 if no supplemental UV LEDs are used in the color measurement instrument. Finally, M3 specifies no UV be included in the illumination, and that the illumination be cross-polarized to remove any effects of specular highlights from measurements. M3 is more commonly used in Europe and for dry-back measurements in offset lithography.
Table 1. Description and primary use of measurement conditions M0, M1, M2, and M3.
Uncontrolled illumination (no other classification)
D50 illumination with controlled UV
UV-cut (UV component blocked)
Paper with fluorescent component removed
UV-cut with cross-polarization
Wet ink measurements in offset
Many newer handheld color measurement spectrophotometers, such as the X-Rite eXact and Techkon SpectroDens, can be configured to measure in multiple measurement conditions. Care should be taken to ensure they are configured properly for the desired application.
Color measurement is based on the complex interaction between the human visual system, illumination, and objects. Every person perceives color differently and there are an infinite number of lighting conditions one might encounter in the world. The goal of colorimetry is to distill this enormous variability into a model that can be used as a single, globally-accepted model for describing the perception of color objects.
This is the second in a series of five articles covering color measurement fundamentals. The next article, “Understanding CIE L*a*b* and ΔE,” discusses the globally accepted model of color perception, the CIE L*a*b* color space, and ΔE, the model for describing the perceived difference between two colors.
Author: CNX OpenStax (CC License, Free to Use) https://commons.wikimedia.org/wiki/File:Figure_36_05_03.jpg
Retina photo by Mark D. Fairchild, Rochester Institute of Technology, found at https://www.cis.rit.edu/fairchild/WhyIsColor/Questions/3-4.html. Eye diagram found at https://www.onlinelabels.com/clip-art/Eye-diagram-without-annotation-123482.htm
 Brain drawing available via creative commons license (no attribution) at https://cdn.pixabay.com/photo/2016/08/18/12/00/brain-1602757_640.png. Eye diagram found at https://openclipart.org/detail/261648/eye-diagram-without-annotation.
Brain drawing available via creative commons license at https://cdn.pixabay.com/photo/2016/08/18/12/00/brain-1602757_640.png.
 A good reference if you want to learn more about this topic, and colorimetry in general, is Billmeyer and Saltzman’s Principles of Color Technology, by Roy Berns (2003). Also, Wikipedia is good starting point.
 Eye diagram found at https://openclipart.org/detail/261648/eye-diagram-without-annotation.
 Illustration of illuminants found at https://www.flexoglobal.com/flexomag/08-September/flexomag-ploumidis.htm, among other links.
AVT Color Measurement Fundamentals Series: