Color histograms further the idea of the luminosity histogram by introducing the primary colors used for display in digital systems which are red, green and blue. In some contexts they are known as RGB histograms because histograms for all three color channels are overlaid on the same histogram.

A color image is rendered by pixels which are themselves mixes of red, green and blue. Besides these, cyan (mix of green and blue; yellow (mix of red and green) and magenta (mix of red and blue) are the colors resulting from the mix or intersection of two primary colors and are appropriately visible on an RGB histogram.

Histograms in most software applications usually include grey histograms which represent all the tones between darks and whites and are indicators of overall perceived darkness or brightness of each pixel. However some software applications such as Adobe Photoshop have the option of display individual color channels on a histogram.


Each color channel represented on its own histogram

Single Channel Histograms

These can come in very handy when editing images which are particularly problematic due to unusual lighting conditions and while the image overall may seem usable, a histogram can show with some accuracy that one or more of the channels is not being rendered with less than acceptable accuracy.

One of the most obvious issues has to do with clipping. In clipping, a cluster of pixels may be so bright that they fail to capture any useful detail. In a pixel (call it RGB pixel for simplicity), any one color may be clipped while others are not. For example an RGB value of 255, 180, 75 indicates that the red channel is clipping while the green and blue channels are not.

singlechannelrgb - Copy

Clipping of the red channel

Still on the subject of clipping in single channels, it is important to understand that clipping does not necessary result in pure white pixels. The formation of pure white pixels would require the red, green and blue channels all to be clipped – which is also known as white point clipping. Yet clipping still manifests itself when considering single colors because the loss of detail can still take place. A pixel can be so highly overexposed in one color, that there is a loss of texture and detail. For these reasons, one of the objectives of checking single channel histograms is to reduce instances of clipping and crushing (where the color is underexposed and is represented with black tones).

Comparing Luminosity & Single Channel Histograms

  • The confusion arises in some cameras and when relying on luminosity histograms alone pixels are considered to be clipping when they have a value of 255 with no regard to the condition of each channel
  • This is because luminosity calculation is done on the overall brightness of a red, green and blue pixels taken as a whole by applying a weighting to each each in the order of 30% for red; 59% for green and 11% for blue
  • Example: A pixel with RGB values of 230, 150 and 60 respectively, the luminance will be given by (200 x 30%) + (150 x 59%) + (60 x 11%) = 155 
  • Clipped red channel: 255, 180, 75, the luminosity will be (255 x 30%) + (180 x 59%) + (65 x 11%) = 191
  • We can describe the red channel in the above example as being clipped, saturated or blown out but the luminosity histogram would not convey this which highlights the importance of single channel histograms
  • We therefore consider luminosity histograms to be inferior to single channel histograms, a mathematical representation which is not suited to color images
%d bloggers like this: