Neural network approach for color matching of ceramic glazes

Aesthetical quality of ceramic tiles is a winning and indispensable requirement in production. Pigment oxides play an important role and more and more sophisticated applications are required in industrial applications. In the present work a neural network model was tested to define the final apparence due to the mixture of ceramic pigment used for wall tiles. Samples were prepared using thirty industrial pigments and one frit for wall tiles. Mixing the pigments, seventy nine samples were prepared in a fast ball mill and applied on tiles. After firing in a industrial cycle, the L*,a* and b* coordinates in CIElab spaces of the surfaces were measured using a spectrophotometer and put in relationship with the pigment compositions to define a model able to calculate the colour when the pigment composition is known. The results show a good efficiency of the color matching algorithm.