3D列印的技術逐漸成熟,然而,彩色3D列印的色域有限,色彩控制不穩定,手工拋光與顯色等成品後加工處理,使得同一批複製品的表面色彩不一致。為了客觀評價3D列印的色彩複製是否準確,本研究提出一套能夠經由彩色影像評估立體器物視覺色差的模式。這個模式包含了影像2D/3D對齊,模擬視覺解析度,計算影像色差,與計算單一色差指標。為了準確拍攝3D物件色彩,本研究提出了一種雙曝光HDR影像融合方法,這種方法使用3階多項式迴歸並且在對數曲線上調整得到接近真實環境亮度之HDR影像。為了準確評估立體器物的視覺色差,本研究分別以實際觀測與HDR影像觀測3D列印色差模型,用以評估適用於立體器物的色差公式加權參數。單色模型的分析中,不論哪一種方式進行觀測評估視覺色差,CIE94與CIE2000色差公式的優化加權參數大約等於(KL, KC, KH)=(2:1:1.5),意味著明暗的差異較不容易區別。多色模型的分析結果表明人型模型膚色的正確的重要性遠大於非膚色區域。影像色差的單一指標,以影像色差的第百分之九十五分位數與視覺色差的關連性最高。
3D printing is widespread in recent years. Color 3D printers are also available in the market now for rapid prototyping and small amount productions. However, evaluating the quality of 3D printing is an important but unresolved issue. The aim of this study is to evaluate color differences of 3D objects based on the HDR image. The experiment established a series of psychophysical experiments to evaluate the visual color differences. Observations includes entity viewing and image viewing. The image includes Multi-exposure LDR photos were integrated as HDR image using 3rd polynomial regression in logarithmic scales. The results were used to develop an imaging system for evaluating color differences of 3D objects. In the analysis of the monochromatic model, no matter which way to observe and evaluate the visual color differences, the weighting parameters of CIE94 and CIE2000 color difference formulae optimization result is approximately equal to (KL, KC, KH) = (2:1:1.5), which suggests that highlight and deep shadows are less important in the perceptual color differences. And the human eye would judge dark color difference easily. In the analysis of the multicolor model, the results indicate that the correct importance of the skin color of the human model. In terms of image differences, the results show that the perceptual color differences are well-correlated to 95th percentile of the image color differences.