Multivariate Gaussian subspatial regression applied to predict the effect of phosphate crystallization aging on the color in silicious conglomerates

作者：Vicente Palacios Victor; Carlos Inigo Adolo; Garcia Talegon Jacinta

来源：Color Research and Application, 2017, 42(6): 799-806.

DOI：10.1002/col.22142

• color difference metrics
• Gaussian process
• multivariate Gaussian subspatial regression
• preservation
• restoration

A new methodology has been applied to the experimental data obtained about a white siliceous conglomerate from Zamora (Spain), which was subjected to 25 cycles of 2 types of aging [freezing/thawing with cooling/heating (T1) and freezing/thawing with cooling/heating+phosphate crystallization (T2)]. Our model (multivariate Gaussian subspatial regression) allows the behavior and prediction of the chromatic coordinates (L*,a*,b*), including more than 25 cycles, to be analyzed. This model is much more flexible than classical models as it allows multiple variable combinations to be predicted in a dynamic way. The final result showed that the conglomerate experiences darkening, yellowing, and reddening, as the number of cycles increase and that the darkening is much less pronounced in T2 due to phosphate crystallization.

• 出版日期2017-12