摘要
开发了一种铭文锦纹样元素的智能提取算法。首先采用双边滤波算法平滑图像的织物纹理;然后对图像做自动色彩均衡处理,提高图片对比清晰度;再设计使用Gra b-Cut算法,确定超像素分割量,有针对性地提取出图案因子;最后,运用Canny算子提取分割后的铭文锦图案因子的轮廓,并利用CDR软件将其矢量优化。对“长乐明光”锦的智能提取实验结果显示,6个图案因子的分割像素准确度PA均大于95%,表明该组合算法可以准确完整地提取出汉代铭文锦纹样的图案因子,为中国铭文锦纹样的再应用提供参考。
This paper designed an intelligent extraction algorithm for inscription-brocades pattern elements was designed.First of all,the bilateral filtering algorithm was used to smooth the fabric texture.Then,the image was processed by ACE to enchance the contrast definition of the image.Next,the Gra b-Cut algorithm was designed.The amount of super-pixel segmentation quantity was determined,and the pattern elements were extracted pertinently.Finally,the contour of inscription-brocades pattern elements was extracted by Canny operator,and CDR software was used to vectorize it.The experimental results of intelligent extraction of“Chang Le Ming Guang”brocade show that the pixel segmentation accuracy PA of six image factors is greater than 95%.It shows that the combined algorithm in this paper can accurately and completely extract inscription-brocades pattern elements in Han Dynasty.This study can provide reference for the reuse of inscriptions in China.
作者
田苗苗
李艳梅
俞凯杰
TIAN Miaomiao;LI Yanmei;YU Kaijie(School of Textiles and Fashion,Shanghai University of Engineering Science,Shanghai,201620,China)
出处
《北京服装学院学报（自然科学版）》
CAS
北大核心
2023年第4期43-49,共7页
Journal of Beijing Institute of Fashion Technology：Natural Science Edition
基金
上海艺术科学规划重大项目(2017wNO.107)。
关键词
汉代铭文锦
特征提取
纹样元素
图像处理
inscription-brocades pattern in Han Dynasty
feature extraction
pattern element
image processing algorithm
