A Case for a Linear Color Difference Metric

Details:

Year: 2015
Pages: 14

Summary:

At TAGA 2008 (Granger,2008), a paper on color differences showed that by adding the influence of the chromatic channels of color vision to the perception of brightness, a set of simple equations could predict color differences as well as the more complex equations of the CIE. This paper further develops the concept where the color space is limited to the sRGB or the ITU-R BT 709-5 color gamut (ISO).

The sRGB or the ITU-R BT 709-5 color standards dominate color reproduction in the communication and reproduction of images. The colorimetric vector space with a defined D65 white point has had successful commercial application to cameras, printers and television. A majority of images are being recorded with these standards. Pointer (Pointer, 1980) defined a gamut of real world colors that is larger than the sRGB color gamut, but success of the sRGB system in the commercial world suggests that on a statistical basis the extreme colors of Pointer's gamut are seldom seen in most of our images. The current paper limits its study of space linearity to the sRGB gamut since the majority of imagery is produced using this standard.

The study of linear transformations over the smaller gamut limit has led to a very simple set of equations that transform RGB values in the sRGB color space. The new QTD color space is shown to be more uniform than either CIE Luv or CIE Lab. This transformation also has the properties of being well correlated to the hue discrimination of the Munsell color space and of being able to predict substrate brightness or whiteness.