Effect of background colour on monitor characterisation

A common solution for transferring images from one device to another without loss of colour fidelity is to characterise each device in terms of CIE tristimulus values. For example, it would be possible with appropriate characterisation procedures to convert the LCD RGB values to CIE XYZ values and vice versa. Characterisation of devices into a standard colour space that is independent of the device reduces the number of transformations which may be required for adequate performance. The first stage in characterisation is to linearise the data termed "gamma correction" for certain devices and then transform the linearised values in CIE XYZ tristimulus values. In order to determine the non-linearity of the characterisation and the matrix for linear conversion between RGB and XYZ it is normal to make colour measurements of certain colour patches displayed on the system. However, it is known that the colour measurements of the patches may vary with the colour and luminance of the background against which they are displayed. Lack of spatial independence is one of the factors that can cause this phenomenon. This raises the question of what the nature of the background should be for an optimal characterisation of a display system. It is likely that what is optimal will depend upon the intended application of the characterised display (for example, is it being used to display simple images in a psychophysical experiment or more complex images in some other setting). This research considers characterisation with four background conditions (white, grey, black and a new Mondrian-like coloured background) and explores the effect of these background effects on the characterisation model’s parameters and on the usefulness of the characterisation in various imaging scenarios.