to be less than unity, i.e. no perfect correlation was
obtained, due possibly to lack of uniformity in the colour
space generated together with some experimental error. At
the same time, R2 values for dark colours were higher than
those for lighter colours.
In view of the similar performance of both CMC and
CIE94 colour-difference equations, it was not readily
apparent which one gave the better performance for
colour acceptance determination and shade sorting.
Consequently, all the data obtained from the 20 individual
sets of colours were illustrated in a single scattergram in
order to assess the overall performance of these equations
throughout the colour space using both linear and
logistical regression. The results of the comparison of R2
are shown in Table 2. Again these results lead to the
conclusion that overall the CMC equation gave the best fit
for generating a uniform microspace for allocating the
colour population in shade sorting.
In addition, the R2 values obtained from the linear
regression analysis were greater than those from logistical
regression using the equations. Linear regression seemed
to give a better fit over the whole batch of samples.