Optimizing color extraction for traditional Yun Jian: a benchmark of clustering algorithms and construction of a heritage color database

This study addresses the lack of systematic algorithmic comparison in digital color extraction of traditional Chinese Yun Jian. It aims to identify the optimal clustering algorithm and construct a structured color database for heritage preservation and design innovation. Three algorithms—K-means, Fuzzy C-Means (FCM), and Gaussian Mixture Model (GMM)—were evaluated on Qing Dynasty Yun Jian patterns using a comprehensive framework measuring execution efficiency, CIELAB ΔEab color difference, and visual segmentation quality. Results showed K-means outperformed others in color restoration accuracy (average ΔEab = 3.65), boundary definition, and adaptability to complex patterns. Using K-means on an expanded sample set, a structured database of twelve representative colors was constructed, with its design utility validated through color spectrum modeling, automatic scheme generation and expert evaluation. This research establishes a clear methodological basis for quantitative color study of traditional costumes and provides a practical data resource for digital archiving and innovative transformation of Yun Jian, bridging cultural preservation with contemporary application.