智動色彩與白平衡校色系統－皮革工藝網站

Leather-Tone Craft Website: Intelligent Color-Balance Calibration System

羅梅君(Mei-Chun Lo) ； 呂泓霖(Hong-Lin Lu) ； 尤勻芊(Yun-Chian You)

《中華印刷科技年報》 2022 (2022/05) Pp. 11-22

摘要

現今台灣的手工皮革是用傳統方式去做手動調色與上色，並以零售方式採買現成的色皮；假使以植鞣皮革加上自行“調色刷染”創作，就常會因受限於購買的顏料色彩不夠豐富，導致無法滿足購買者在美感設計上“多樣化”的喜好需求。所以，對於有心想要自行生產製作，並將產品商業化的銷售或製造商家而言，勢必要尋求比較科學化的方法，以解決「效率缺乏」、「選色不易多樣化」、「刷染顏色不易管理和控制」以及「穩定性不足」等問題。為此，本研究擬將「色彩科技度量學」（Colorimetry，即數位色彩管理科學）應用到傳統的手工皮革調色工藝中；首先，將嘗試（1）「建立皮革顏色四原色＋水（C,M,Y,K,W）色彩比例調色的數據化演色表資料庫」，並（2）「利用深度學習的技術，推演出屬於“頻譜型”─可克服“條件等色”（Metamerism）問題的配色演算法」；隨後，（3）「運用此研發出的光譜演算法，即可根據使用者所需求的標定顏色，查尋到最佳的原色墨調配比例。最終的目標，將實踐（4）「皮革工藝網站的開發」，除了以網頁呈現染色後的皮革2D/3D作品樣貌之外，也提供後端頻譜型的配色演算法服務，以便讓相關或有興趣的皮革商家也能迅速的找到其皮革顏色的光譜數據（和色彩數據），作為選取調配適當比例的原色顏料進行染色之用。因此，本研究發現將機器學習與深度學習結合在傳統手工染色產業上能夠有一大突破，透過本研究所開發的系統雛型能夠將科技更進階的被廣泛延伸推廣到相關應用領域，例如印刷及紡織業等相關產業。

關鍵字

條件等色 ； 多頻光譜演算法 ； 深度學習 ； 灰平衡演算法 ； 色料CMYK色彩模組

並列摘要

Today's hand-made leather in Taiwan is manually colored in the traditional way, and ready-made skins are purchased retailly. If you create with vegetable tanning leather and plus with your own ＂color grading brush dyeing＂, it is often limited by the lack of color of the pigments purchased, and resulting in the inability to meet the buyer's preferences for ＂diversification＂ in aesthetic design. Therefore, for the sales or manufacturers who want to produce their own production and commercialize the product, it is necessary to seek a more scientific method to solve the problems of 1) ＂lack of efficiency＂, 2) ＂color selection is not easy to diversify＂, 3) ＂brush and dye color is not easy to manage and control＂ and 4) ＂lack of stability＂. To this end, this study proposes to combine ＂Colorimetry＂ (the science of digital color management) applied to the traditional hand-made leather grading process. First, we tried 1) ＂establishing a color-proportion grading database, based on color tables (i.e. database) of four primary colors of leather pigments plus water (i.e. C, M, Y, K, W) ＂, and 2) ＂ using deep-learning techniques to deduce color-matching algorithms in terms of the ＂spectrum basics＂ that can overcome the conditional isochromatic (Metamerism) problem. Subsequently, 3) ＂Using the spectral algorithm developed here, the optimal mixing ratio of primary color inks in question can be found according to the resulted color-appearance of calibration-color, required by the user＂. The ultimate goal in this research is to practice 4) ＂the development of leather-tone craft website＂. In addition to presenting the appearance of dyed leather 2D/3D works on the web page, the developed web-site but also would provide back-end services using the spectrum color-matching algorithm derived. It could allow related or interested leather merchants quickly find or allocate appropriately spectral-based proportions of primary color pigments for dyeing their requested leather color. Satisfactorily, this research had a major breakthrough in the traditional hand-dyed industry. The prototype of the developed system could be further extended to related application fields, such as printing and textile industry.

並列關鍵字

Metamerism ； Multispectral Algorithm ； Deep Learning ； White Balancing ； CMYK Color Mode

參考文獻

Bruce Justin Lindbloom (2017), RGB Working Space Information, (2022/03/25， 取自 : http://www.brucelindbloom.com/ )

Dobilas, S. (2020), Machine Learning: Support Vector Regression (SVR) — One of the Most Flexible Yet Robust Prediction Algorithms, Towards Data Science. (2022/03/25， 取自 : https://towardsdatascience.com/support-vector-regression-svr-one-of-themost-flexible-yet-robust-prediction-algorithms-4d25fbdaca60 )

Kang, H. R. (1997), Color Technology for Electronic Imaging Devices, SPIE Optical Engineering Press.

Lee, H. C. (2005), Introduction to Color Imaging Science, Cambridge University Press. p. 450. ISBN 978-0-521-84388-1.

Maloney, L. T. and Wandell, B. A. (1986), Color Constancy: A Method for Recovering Surface Spectral Reflectance, Vol. 3, No.1/January 1986, J. Opt. Soc. Am.