About me

Course Work



CSE300 Multimedia Systems: Student Work

Basic Image Properties and Color Coordinate Systems

Color only has meaning when we talk about biological vision. Visible light is a small range of frequencies of electro-magnetic radiation where different frequencies correspond to different color sensations.

Intensity is related to the power of the emitted electromagnetic radiation. As Poynton indicates, this is best measured in watts per square meter. Our vision perceives intensity as brighter or darker sensations.

As defined by the CIE, brightness is the attribute of a visual sensation according to which an aera appears to emit more or less light.[2]

Luminance is the perceived intensity of a source, which is non-linear in our vision. As Poynton indicates, this is best measured in candelas per meter squared.

Since our vision is non-linear, it does not make sense to always use fixed quantities like watts per square meter. Lightness attempts to quantify what we see. For example, a source that is 50% lighter usually will not have a luminance that is 50% greater.

As defined by the CIE, hue is the attribute of a visual sensation according to which an area appears to be similar to one of the perceived colours, red, yellow, green and blue, or a combination of two of them.[3] If you view an image through a colored filter, the image will appear to take on the color of the filter. This color will be the hue of the image.

Saturation can be refered to as the richness of the color. If a color appears dull or grey, then it has a low saturation. A color that appears very full compared with the rest of the image has a high saturation.

As Paynton describes chromaticity, it is often convenient to discuss "pure" color in the absence of brightness.[1]

White Reference
If you turn on two lights in a room, one tungston and one flourscent, you will notice that the first light appears yellowish and the other one appears blue-greenish. However, if you only turn on one light, you will say that each one is white. There is no physical definition for white, so it is important to specify a white reference and then base all other colors off of that reference point. Human vision will attempt to adapt/accept a given white reference provided that there is only one white reference (i.e. only one light on).

Color Temperature
When many solids are heated, they begin to emit broad spectrum radiation. As the temperature increases, the radiation gains in intensity and the range of frequenies increases. It is possible to provide a reasonable specification for a white reference point by the temperature given in degrees Kelvin. A typical white point for computer monitors is 9300K.

RGB - Red Green Blue
Since human vision appears to be based on three types of receptors, each sensitive to a different group of wavelengths, it makes sense to try to describe colors using this model. Our eyes have different areas that are particularly sensitive to red, green, and blue light. By mixing these stimuli, we can generate most of the visible spectrum. This is an additive system which closely approximates how a CRT monitor generates its colors.

CMY - Cyan Magenta Yellow
In a subtractive system like CMY, colors are defined by how much color is absobored between the source and your eye. For example, if you pass white light through a gas, you may see a color. This is because some wavelengths are absorbed by the gas. This is a subtractive system which is approximated by the CMY system. The real usefulness of this is with printing since paints tend to mix in a subtractive manner. By mixing all three colors, you get black.

CMYK - Cyan Magenta Yellow Black
As with CMY, CMYK is a subtractive system. The K (black) is used to reduce the cost of printing (one ink verses three), to avoid strict registration requirements (slightly offset plates for CMY will produce highly visible colors around black text), and to provide a "true" black (mixing CMY usually gives a very dark, muddy grey).

HSI, HSB, HSL - Hue Saturation (Intensity, Brightness, Lightness)
These systems were created to allow a user a simple way of specifying a color. The basic color can be defined as the hue. If the color should be rich, a high saturation value can be used, or if a grey shade is needed, the saturation can be reduced. The lightness allows the user to push the color toward a brighter shade of the color. As Poynton points out, the typical HSB calculations introduce visible discontinuities in color space.[1]

YPbPr & YCbCr
The YPbPr system is designed for analog values in the range [0..1], whereas YCbCr is designed for digital values in the range [0..255]. According to Poynton, vision has poor response to spatial detail in colored areas of the same luminance, compared to its response to luminance spatial detail.[1] Therefore, it may be advantageous to remove the luminance information and use the remaining bits to transmit the color information.

The XYZ system uses two spectral weighting curves (X and Z) and the luminance (Y) to define colors. As Poynton indicates, the components are proportional to physical energy, but their spectral composition corresponds to the color matching characteristics of human vision.[1]

In an attempt to create a color system that is perceptually uniform, the CIE defined two standards. These are CIELAB and CIELUV. By perceptually uniform, I mean that a change to one of the values will be equally noticible regardless of its initial value. The example that Poynton uses is a volume knob on a stereo. Human hearing is logarithmic in nature, so the volume knob is also logarithmic. When we turn the knob we perceive a linear change in the volume even though the sound is increasing at a logarithmic rate. The problem with the CIELAB and CIELUV systems is that they are to intense computationally for video systems at this point.

R'G'B' - non-linear Red Green Blue
Since human vision is non-linear, it makes sense that a color system should try to approximate this to become closer to perceptually uniform. Standard weights have been assigned to RGB by the CIE so that the system is closer to being perceptually uniform. While not perfect, this system requires significantly less processing power and is commonly found in video systems.

[1] Poynton, Charles. Frequently Asked Questions about Color.
[2] Poynton, Charles. Frequently Asked Questions about Gamma.
[3] Colour Science Glossary.
Last Modified: December 13, 1998 - Barry E. Mapen