ICC specification prescribesthe types of ICC profiles, mandatory and optional data structures and transformation models. On the other hand, the process of creating an ICC profile is undefined in order to open space for improvement and optimization, depending on the use of ICC profiles. Selection of characterization models that will be used for populating profile`s data structures, as well as method of their use is left to the development team. It is known that the accuracy of the profile is affected by factors that we can control and those we can not control. Some of the factors that we can control are the choice of characterization models, the size of the transformation table and the correct use of one-dimensional "A" curves and "B" curves. In this work the accuracy of profiles made with different characterization models will be compared. Interpolation tables, regression model and polyharmonic spline model will be used.. The research will determine the quality of fit and predictive power of different models in one process. Some of the earlier studies showed similar performances of regression and spline method. It is assumed that the regression as fitting method can have the advantages when larger measurement error or variability of the process is present, but that the spline as interpolation method is more accurate.