Abstract:
Most printed labels required customer attention. Consequently, they employ extra or special colors apart from 4-color process, CMYK. These special colors need being matched with the standard colors which are given by brand owner or designer beforehand. The algorithm of color matching process usually includes iterative method. The goals of the process are accuracy, precision and high speed. This research is aimed to compare the efficiency of UV flexographic ink color match prediction on polypropylene label using artificial neutral network (ANN) and color match prediction (CMP) software. The Multilayer perceptron ANN ; MLPANN and X-Rite ink formulation software were employed. Preparation of ink database for both methods were different. However, the number of primary colors or inks used for comparison were the same. For the CMP method, 8 levels of ink concentration including 100% ink were made by diluting ink with medium into 7 levels and printed on polypropylene label. Their spectral reflectances were measured and the absorption and scattering coefficients were subsequently calculated for all concentrations. For the MLPANN method, 2 of primary inks were mixed to obtain 159 secondary mixtures and 3 of primary inks were mixed to obtain 105 tertiary mixtures. They were printed on to the same substrate. Again their spectral reflectances were recorded. Eighty percentage of database then were randomly selected for training set. The number of hidden units in hidden layer were optimized. In the CMP process using software and single-output ANN method, 15 color test samples were used and CIE color differences (ΔE*94), between samples and predicted formulae were used as one of evaluation criteria in terms of accuracy apart from speed of calculation, time of making database and cost. We found that the vantage of the MLPANN is accuracy with less average ΔE*94, whereas the disadvantages are time consuming in making database which consequently the labour cost is high. Speed of calculation is approximately the same.