Abstract: Chaoyuan Painting is a pinnacle of Taoist murals from the Yuan Dynasty, showcasing the beauty of traditional Chinese colors through its vivid and magnificent color usage. The unique and stylish colors of its costumes can provide rich inspiration for national fashion design. However, the current research on the color of costumes in Chaoyuan Painting lacks support from scientific theory. This study focuses on the costumes of 260 immortals in Chaoyuan Painting. Firstly, a sample library was established to extract color and matching rules, based on which a color network model was created. The Munsell color system was then used to analyze color features. Finally, the color network model was applied to the design process of yoga clothes by using virtual fitting software to present the design effects.
To obtain the node values and color proportions of the color network, a costume sample library of Chaoyuan Painting was established for color extraction. 260 high-quality images were selected from the 290 images of immortals in Chaoyuan Painting, of which the non-clothing parts were removed and noise elimination and texture smoothing were performed. Each costume image was clustered for the first time to obtain corresponding 8-color color board . Then the secondary clustering was applied to male and female costume images, and 16 colors for each gender and the color proportion information were acquired. To establish color network node connections, the Apriori algorithm was used to extract color-matching rules. Binary and multiple matching rules were extracted with a minimum support of 0.3 and a minimum confidence of 0.7. Finally, separate color network models for male and female costumes of Chaoyuan Painting were established based on the above results. 
Based on the color network, the Munsell system was employed to analyze the color features of costumes in Chaoyuan Painting. Munsell labels were used to notate the extracted colors to objectively describe their color attributes. Male and female costume colors were then represented by Munsell color spectra, including Munsell hue and Munsell chroma. The results showed that the costumes of male and female immortals in Chaoyuan Painting were divided into red, yellow-red, yellow, yellow-green, blue-green, and achromatic color series. The red series and blue-green series were generally distributed in the medium saturation and medium lightness range, while the yellow-red, yellow, and yellow-green series were distributed in low saturation with a large lightness span. Finally, the costume color combination in Chaoyuan Painting was analyzed according to the Munsell’s theory of color harmony, revealing a predominance of cool-warm and light-dark contrasts.
The color network model for male clothing was applied to the design of yoga clothes, and was presented by the virtual fitting software Style3D. Following the Munsell’s theory of color harmony, three color-matching rules were chosen from the blue-green color with largest proportion. Colors were assigned to various designs to create preliminary plans, followed by adjustments in color area and texture to achieve final effects. At last, the fuzzy comprehensive evaluation method was used to evaluate the color application scheme which showed good results. This study provides theoretical support and case references for the objective analysis of costume colors in Chaoyuan Painting and their application in fashion design.

Key words: color network, K-means clustering algorithm, costumes in Chaoyuan Painting, Munsell color theory

摘要： 为了客观地探究和反映元代壁画《朝元图》服饰的色彩设计思维，以260位神仙的服饰为研究对象，对其色彩特征和搭配关系进行量化研究。首先逐一去除每位神仙图像中的非服饰部分，使用中值滤波降噪平滑，建立样本库。再使用K-means聚类算法分别对男性和女性服饰进行色彩提取，使用Apriori关联规则算法挖掘色彩搭配规则，并以色彩网络模型呈现二元色彩搭配情况。然后基于孟塞尔色彩体系理论，进行色彩属性及色彩调和方式分析，发现《朝元图》男女服饰色彩都分为6种色系，色彩分布规律一致，色彩搭配多使用冷暖对比与明暗对比的调和。最后依据所得色彩网络模型，结合孟塞尔色彩理论挑选色彩搭配规则，进行瑜伽服色彩设计。研究结果可为现代服饰色彩的设计应用提供参考。

关键词: 色彩网络, K-means聚类算法, 《朝元图》服饰, 孟塞尔色彩理论