摘要
社会的快速发展为少数民族文化的发展创造了崭新的环境。当前对研究对象的配色方案中,大多是对色彩本身进行研究而鲜有结合人的主观意象,为挖掘意象词与服饰配色之间的隐含关系,建立用户需求意象与服饰色彩构成及占比之间的映射模型,提出了一种基于用户需求意象的多重K-means-ELM侗锦配色方案。首先,通过对侗族织锦色彩样本的提取,运用语意差异法(SD法)获得用户的认知数据源。结合聚类分析对侗族织锦图像进行分类和色彩提取。其次,基于色彩提取结果采用K-means对相似色彩构成的图像进行多重聚类划分得到具有相似色彩构成及占比的图像集合。最后,以侗族织锦色彩样本、色彩比例关系和HSV空间数值等信息与用户意象建立映射关系,采用极限学习机(Extreme learning machine,ELM)对多重划分得到的聚类图像样本单独建模,构建用户需求意象的多重K-means-ELM侗锦配色模型进行侗族服饰的配色研究。经过算例分析,将配色模型运用到以“星河”为主题的现代插画设计实际创作中进行验证,得出了该模型的可行性与科学性,所提出的配色方法提高了建模精度、降低了配色难度,从而细化意象—色彩之间的映射关系。
The rapid development of society has created a new environment for the development of minority cultures.Most of the color schemes of the current research object focus on the color itself and rarely combine the subjective image of human.To explore the implicit relationship between imagery words and clothing color schemes,and establish a mapping model between user demand imagery and clothing color composition and proportion,a multiple K-means-ELM Dong brocade color scheme based on user requirement imagery is proposed.Firstly,by extracting the color samples of Dong brocade,the semantic difference method(SD method) is applied to obtain the cognitive data source of users,and the cluster analysis is used to classify and extract the color of the Dong brocade images.Secondly,based on the color extraction results,K-means is used to divide images with similar color composition into multiple clusters to obtain a collection of images with similar color composition and proportion.Finally,the mapping relationship between Dong brocade color sample,color proportion relation,HSV spatial value and user image is established,the extreme learning machine ELM is used to model the clustered image samples obtained by multiple divisions,and a multiple K-means-ELM color matching model of user demand image is constructed to study the color matching of Dong clothing.After the analysis of calculation examples,the color matching model is applied to the actual creation of modern illustration design with the theme of "Galaxy" for verification,which verifies the feasibility and scientificity of the model.The proposed color matching method improves the modeling accuracy and reduces the difficulty of color matching,thus refining the mapping relationship between image and color.
作者
张超
闫茹玉
朱晓君
ZHANG Chao;YAN Ruyu;ZHU Xiaojun(Art College,Guizhou University,Guiyang 550003,China)
出处
《丝绸》
CAS
CSCD
北大核心
2022年第12期108-118,共11页
Journal of Silk
基金
贵州省本科教学内容和课程体系改革项目(2021023)
国家社会科学艺术学项目(22BG138)
贵州省研究生科研基金项目(黔教合YJSKYJJ[2021]051)
贵州大学研究生创新人才计划项目(贵研CXRC202108)
贵州大学高等教育研究项目(GDGJYJ2021007)。