Colour quality of facial prostheses in additive manufacturing

Abstract

Recent progress in additive manufacturing technology has improved the realistic colour reproduction of 3D facial prostheses with computational optimisation of skin colour profiles. The colour appearance of the prosthetic surface depends on both the spectral characteristics of the surfaces and the scene illumination. Considering everyday environments, the colours of prosthetic surfaces should appear constant under various illuminations, although such evaluations on facial prostheses have had limited success to date. In this study, colour quality was assessed throughout the additive manufacturing process, namely, from the colour profile optimisation to the colour reproduction on the 3D facial prostheses. Colour profiles optimised for typical skin colour samples were applied to manufacture facial prostheses with two skin types, Caucasian and Chinese. The colour quality was assessed by the colour difference metric CIEDE2000 and spectral similarity against corresponding real skin data. The constant colour appearance of the prosthetic surfaces under different illuminations was estimated by introducing a reproduction colour-constancy index. The CIEDE2000 between the prosthetic and real skins was approximately 7.2 on average over all skin types and facial areas, which is slightly larger than the acceptable perceptual error. The level was relatively constant under different illuminations selected from the CIE standard daylight and fluorescent lights. The reproduction colour-constancy index ranged from 0.34 to 0.89, which is remarkably similar to the level observed in traditional colour constancy data in vision sciences. Spectral errors were close to those obtained by computational spectral reconstruction from digital RGB colours. These results suggest that the proposed colour management for facial prostheses may satisfy the requirement of colour quality in everyday environments with various illuminations. The causes and further improvement of the remaining errors in the manufacturing process are also discussed.