Colour Characterisation of LCD Display Systems

The main purpose of this research is to study the colour characterisation of digital display systems. Three distinct models for characterisation (GOG, PLCC and PLVC) are evaluated and compared and for two of these models (GOG and PLCC) two different sets of linearisation samples (either colour-ramps or grey-ramp samples) are used to perform the linearisation. To evaluate these models’ colorimetric measurements are made for 20 different display devices and colour characterization performance is reported as the main measure. Characterisation performance is calculated using several sets of samples including the widely used Macbeth ColorChecker chart and two new charts called Chart4 and Matlab60 (one of which was based on a method previously published by Cheung and Westland and another was based on a new method).
A key aspect of this work is that all 256 levels of intensity were measured for the colour-ramps and for the grey-ramp linearisation samples for each of the 20 displays to allow subsampling of these data to explore the effect of the number of linearisation samples on characterisation performance. When the number of linearisation samples used was small (less than 10) the GOG model sometimes resulted in the smallest characterisation colour differences. However, for the PLCC and PLVC models performance tended to increase with the number of linearization samples and both of these models outperformed GOG with more 10 linearisation samples. For the PLCC model, better performance was usually obtained using the grey-ramp linearisation samples rather than using the colour-ramps linearization samples. It was possible, for each of the 20 displays, to reach average ab values that are less than 1.5 (ab <1.5, 90%) or ab < 1.0 (75%); however, the model that yields the best performance is difficult to ascertain in advance (a good strategy would be to evaluate all five models and select the one that performs best for the characterisation of any particular display). However, in the majority of cases, lowest colour differences (ab) were obtained using the PLCC model and all 256 of the grey-ramp samples for linearisation. This work has compared the performance of five different models using a large number of displays and has allowed a number of recommendations to be made about display characterisation. Although the majority of the work in this thesis was based on stationary displays the effect of motion on characterization performance was also explored. This is important since moving images are now commonplace in many applications. The results showed that a moving background has a small, but statistically significant, effect on the colour of patches.