Not sure, but maybe you could test by showing people two similar colors and asking them to decide if the are identical.
With enough samples (and enough individual participants) you establish how many "statistically-significant distinguishable differences" exist along a given path through the color-space.
Then combine those and get a sort of topographic grid...