Here are some interesting examples which inspired the
investigation direction. We created the test images by repeating a
2x2 matrix, which gives the same local mean values in
CMY(K). We are expected to get very similar colors, but..
| local pattern (cmyk) | microscope version | scanned image |
| (1000) (0100)
(0000) (0010) |
||
| (1010) (0100)
(0000) (0000) |
||
| (0000) (0000)
(1100) (0010) |
There are several apparent reasons for this, such as poor
registration, dot size variance, etc. However, we believe that there
are some more fundamental reasons behind, i.e., color interaction
in spatial and spectrum domain.