基于图像的织物图案和颜色主题提取研究进展  

Research progress of image-based fabric pattern and color theme extraction

作　　者：赵崧灵 徐凯忆 张健[1] 顾冰菲 ZHAO Songling;XU Kaiyi;ZHANG Jian;GU Bingfei(School of Fashion Design&Engineering,Zhejiang Sci-Tech University,Hangzhou 310018,China;Zhejiang Provincial Research Center of Fashion Engineering Technology,Zhejiang Sci-Tech University,Hangzhou 310018,China;Key Laboratory of Silk Culture Heritageand Products Design Technology,Ministry of Culture and Tourism,Zhejiang Sci-Tech University,Hangzhou 310018,China)

机构地区：[1]浙江理工大学服装学院,杭州310018 [2]浙江理工大学浙江省服装工程技术研究中心,杭州310018 [3]浙江理工大学丝绸文化传承与产品设计数字化技术文化和旅游部重点实验室,杭州310018

出　　处：《丝绸》2022年第8期90-99,共10页Journal of Silk

基　　金：国家自然科学基金项目(61702461,61702460);中国纺织工业联合会科技指导性项目(2018079);中国纺织工业联合会应用基础研究项目(J202007);浙江理工大学科研业务费专项资金资助项目(2020Q051);浙江理工大学服装服饰文化创新团队项目(11310031282006)。

摘　　要：为提高织物设计的效率及多样性,文章从“再设计”的角度梳理“织物-织物”的设计流程,重点整理织物图案形状及颜色主题的主要提取方法。大部分提取研究主要利用织物图像底层视觉特征中的形状及颜色特征,其中针对图案形状的提取,大多采用边缘检测、阈值、聚类、区域及其他特定理论对图案轮廓或整体区域进行分割;针对颜色主题的提取,则通过聚类、中位切分及八叉树等颜色量化方法建立和谐的配色规律,两个方向的提取算法可以相互借鉴。通过图像实例对不同提取方法的特点进行分析比较,发现单一算法具有一定的局限性,采用多特征、多算法融合的方式可有效提高提取的精度和效率。最后对织物再设计流程的完善及其在纺织服装领域的应用发展进行展望。In recent years,many apparel brands have used digital means to realize the consumers’online design of personalized garments.Fabric is an important factor for garment design,and the pattern and color matching are indispensable elements as important visual features.The traditional manual design is complicated,and therefore,it is an effective way to design with computer technology to extract appropriate elements from fabric images.However,there are some problems with existing applications of pattern and color extraction,including imprecise automatic segmentation,complicated manual operation and irrelevant subsequent applications,making the whole design process inconsistent.Therefore,how to improve the accuracy of pattern segmentation and color-theme extraction needs to be explored and optimized by scholars,which can be applied in practical production more efficiently.In this paper,from the perspective of"fabric redesign",the"fabric-fabric"design process is briefly proposed to provide a new way for fabric pattern design.The redesign process is summarized into four steps,including image pre-processing,fabric pattern extraction,color theme extraction,and design re-use.The main extraction methods of fabric patterns and color themes are concluded,and the different algorithms are compared through examples.The main methods of fabric pattern extraction are edge detection,threshold segmentation,clustering segmentation,region-based segmentation,and other theory-specific segmentation methods.With the development,deep-learning optimization algorithms are combined into the pattern extraction algorithm.Through the analysis of the examples,the extraction result of each basic algorithm is different,with similar problem of missing information on patterns with small areas.In general,the appropriate extraction algorithms need to be chosen or improved according to different fabrics or application purposes.As for color-theme extraction,the method of color quantization is often adopted,which can be divided into two ways based on freque

关 键 词：图像处理 图案分割 颜色主题 特征提取 颜色量化 织物再设计 

分 类 号：TS941.26[轻工技术与工程—服装设计与工程]