‘Mura’ is a type of defect caused by irregular luminosity variation in liquid crystal displays. It is currently inspected and graded against neutral grey backgrounds in factories by well-trained human assessors. The aim of this research is to provide more reliable grading results from experts in agreement with ordinary observers watching real complex images on LCDs. This work is a pilot study. Since colour is one of the essential elements of complex images, we initially chose to focus on the influence of uniform coloured backgrounds on Mura detection. Analysis of the results shows that both CIELAB and CIEDE2000 colour-difference formulae are useful tools in establishing just-noticeable difference (JND) criteria for Mura grading tasks. The small correlation coefficients between the predicted and visual results, however, imply better experimental techniques should be applied in the future study. The new experiments will be conducted to take into account the spatial backgrounds.