當使用者在挑選化妝品時，可能會花費很多時間在繁雜的試用過程。如果可以先用數位模擬彩妝的方式，提供一個範例影像，讓使用者觀看本身畫上這些妝時的外貌，就可以省去使用者在現場反覆試用的時間，以及化妝品不斷在臉上塗抹再卸妝時造成的傷害，完成化妝商品的挑選。
為了達成此目的，本論文提出一套完整的數位模擬彩妝流程，它可以讓一張未化妝的人臉影像，以另一張已化妝的人臉影像，模擬出近似的彩妝效果。首先使用一種特徵點擷取演算法，可以擷取人臉輪廓上的特徵點。接著，將已化妝影像的臉部輪廓，變形至與未化妝影像上類似的輪廓位置。在影像色彩的處理上，從原始的8-bit RGB訊號轉換至CIELAB色彩空間，分解出明度層與色彩層，接者使用保留邊界的模糊方式，將明度層與模糊後的結果相減，可以得到細節明度層，而模糊結果為模糊明度層。最後經過一系列的影像處理，再由CIELAB轉換回8-bit RGB訊號，完成數位彩妝之模擬。
確立了數位彩妝模擬的流程後，本論文調整了明度和彩度相關之係數，生成出彩妝效果介於未化妝影像與已化妝影像間的三張影像。接著用特徵點選出臉頰的一個區塊，計算其平均CIELAB色度值，再計算模擬影像與已化妝影像間的色差值。從結果可以發現，當係數愈大時，模擬出來的結果，不論在明度或彩度的數值上，皆會愈靠近已化妝之影像，證實模擬的結果，確實達到模擬彩妝之目的。

When users select cosmetics, they may spend much time in the process of trying the products. If we can provide a series of template images, they can select the cosmetic effects on their face to do a simulation result using digital facial makeup technology. The procedure can save the time of trying cosmetic and reduce the damage caused by makeup and cleaning up process to help the user choose the proper cosmetics.
This study presents a comprehensive digital makeup process to achieve this goal. It can simulate a similar cosmetic effect on a user’s facial image from the original image with sample makeup images. At the beginning, we captured the feature points on the contour of face. The selecting part of the face with makeup was morphed into the other one who was without makeup. In the processing of color transformation, the 8-bit RGB signals was converted to the CIELAB color space, which includes the lightness channel and the color channels. With edge preserving smooth method, the detailed lightness layer was processed from subtracting lightness channel with smoothed image. And the smoothed image can be regarded as the blurred lightness layer. Finally, through a series of image fusion processing, and converting the CIELAB signals back to 8-bit RGB signals, digital facial makeup was completed.
Ensured the digital makeup process, the paper adjusted the parameters related with lightness and chroma to generate 3 simulated images whose cosmetic effect were between original image and sample makeup image. Then, the image of a part of cheek, which was selected by the feature points of face, was picked up to calculate the average CIELAB values and the color difference between the original image and simulated images. The simulated image is as near to sample makeup image in both lightness and chroma as the parameter become larger. It proved that the result achieved the purpose of cosmetic simulation truly.

[1] Howard Anton, 2005, “Elementary Linear Algebra, 9th Edition “, Warps And Morphs, New York: John Wiley & Sons.
[2] Z. Farbman, R. Fattal, D. Lischinski, and R. Szeliski. “Edge- preserving decompositions for multi-scale tone and detail manipulation.” ACM Trans. Graphics, 27(3):1–10, 2008. 

[3] Akshay Asthana, Stefanos Zafeiriou, Shiyang Cheng and Maja Pantic. “Robust Discriminative Response Map Fitting with Constrained Local Models.” In Proc. of 2013 IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2013), Portland, Oregon, USA, June 2013.
[4] Hung-Shing Chen, Shih-Han Chen, Yen-Hsiang Chao, Ronnier Luo and Pei-Li Sun, “Applying Image-based Color Palette for Achieving High Image Quality for Displays.” Color Research and Application, Volume 39, Issue 2, pp. 154-168. 2014.
[5] 吳欣穎,「不不同環境照度度下液晶示螢幕膚色再現之研究」, 碩士論文, 世新大學, 台北 (2006)
[6] 許友嘉,「膚色相關色空間下影像相依性膚色修正方法」, 碩士論文, 國立立臺灣科技大學, 台北 (2010)
[7] W.-S. Tong, C.-K. Tang, M. S. Brown, and Y.-Q. Xu. “Example-based cosmetic transfer.” Computer Graphics and Applications, Pacific Conference on, 0:211–218, 2007. 

[8] D. Guo and T. Sim. “ Digital face makeup by example.” In 
Proc. of IEEE Conf. on Computer Vision and Pattern Recog- 
nition (CVPR), 2009. 

[9] A. Woodland and F. Labrosse. “On the separation of luminance from colour in images.“ In International Conference on Vision, Video and Graphics, Edinburgh, UK, 2005. The Eurographics Association.
[10] 林新強,「博物館典藏繪畫之色彩品質評價」, 碩士論文, 國立立臺灣科技大學, 台北 (2014)
[11] F. Nars. “Makeup your Mind”. PowerHouse Books, 2004. 

[12] Rafael C. Gonzalez, Richard E. Woods, 2007, “Digital Image Processing” Prentice Hall.
[13] 黃日鋒、詹文鑫、陳鴻興、胡國瑞、徐道義、孫沛立、羅梅君 編著, 陳鴻興 編審, “顯示色彩工程學”, 全華圖書股份有限公司, 台北, 2011, 第139頁。
[14] M. R. Luo, G. Cui, and B. Rigg. “The Development of CIE2000 Colour-Difference Formula: CIEDE2000.” Color Research and Application 26:5 340-350, 2001
[15] Microsoft: How old do I look?, 網址：https://how-old.net/, 上網日期：2015-05-07
[16] Ohta, Noboru, and Alan Robertson, Colorimetry: fundamentals and applications, (John Wiley & Sons, 2006).
[17] Melgosa, M., Quesada, J. J.; Hita, E. “Uniformity of some recent color metrics tested with an accurate color-difference tolerance dataset”. Applied Optics 33 (34): 8069–8077. doi:10.1364/AO.33.008069. 1994.
[18] 朱正生 編著, “彩妝密碼2：朱正生教妳四大彩妝攻略”, 時報文化出版企業股份有限公司, 台北, 2008