| 摘要 |
PROBLEM TO BE SOLVED: To simulate looks of colors under a desired light source by converting a beam splitting distribution, to which color filtering processing is applied, into three feature parameter coefficient signals and converting three feature parameter signals into color separation signals. SOLUTION: The parameters of converting parts 1 and 5 on a neural network and the color filter of a converting part 3 are read out of a storage part 7 and set to the neural network of the converting parts 1 and 5 and the matrix of the converting part 3. Then, the color separation value signal of a source image is read from an input part 6, the color separation value signal is converted into three main component vector coefficient signals by the converting part 1, and these respective main component vector coefficient signals are converted into a beam splitting distribution signal by the converting part 2. The beam splitting distribution signal is converted into three main component vector coefficients by a converting part 4 and these coefficients are converted into color separation values by a converting part 5. On the basis of these color separation signals, an image estimating the looks of image colors is outputted by an output part 8.
|
