| 摘要 |
<P>PROBLEM TO BE SOLVED: To accurately predict a printing result through less printing/measurement. <P>SOLUTION: Printing of a color patch and measurement of spectral reflectance R(λ) are performed as to only a small number of first representative points selected from grating points generated in an ink amount space. In this case, grating points are selected as first representative points without generating any deviation in the ink amount space. Then spectral reflectance R(λ) at second representative points is predicted through a neural network NNR structured on the basis of the first representative pints as learning data CD1. Consequently, obtained is spectral reflectance R(λ) at nodes (the first representative points plus the second representative points) which are enough to predict spectral reflectance R(λ) by a cell-division Yule-Nielsen spectral Neugebauer model. <P>COPYRIGHT: (C)2009,JPO&INPIT |
