纺织学报 ›› 2020, Vol. 41 ›› Issue (08): 145-151.doi: 10. 13475/ j.fzxb.20190806507

• 综合述评 • 上一篇    下一篇

服装款式图识别与样板转换技术研究进展

李 涛1, 杜 磊1,2, 黄振华1, 蒋玉萍1, 邹奉元1,2   

  1. 1. 浙江理工大学服装学院, 浙江杭州 310018;
    2. 浙江理工大学浙江省服装工程技术研究中心, 浙江杭州 310018
  • 收稿日期:2019-08-26 修回日期:2020-04-09 出版日期:2020-08-15 发布日期:2020-08-21

Review on pattern conversion technology based on garment flat recognition

LI Tao1, DU Lei1,2, HUANG Zhenhua1, JIANG Yuping1, ZOU Fengyuan1,2   

  1. 1. School of Fashion Design & Engineering, Zhejiang Sci-Tech University, Hangzhou, Zhejiang 310018, China;
    2. Apparel Engineering Research Center of Zhejiang Province, Zhejiang Sci-Tech University, Hangzhou, Zhejiang 310018, China
  • Received:2019-08-26 Revised:2020-04-09 Online:2020-08-15 Published:2020-08-21

摘要:

为揭示款式图与服装样板之间的转换机制,概述了依据款式图进行人工制板的过程和方法,阐述了服装款式图特征参数识别和机器学习识别2 种识别方法,重点论述在款式图识别基础上的样板转换技术,并对其优缺点进行了分析。参数化和匹配转换是目前最常用的样板转换方法:参数化转换适用于款式结构较为定型的服装,转换精度高,但不同款式图需要建立各自的转换模型;匹配转换可实现样板的快速转换,鲁棒性高,且规避了样板设计规则,不足是精度较低,且前期需要构建庞大的数据集作为训练集。研究认为,未来可从细化款式图识别粒度、服装款式图面料参数样板多领域跨域匹配、部件化样板智能生成3 个领域开展相关研究。

关键词: 服装款式图, 样板转换, 图像识别, 部件化样板, 参数化转换, 匹配转换

Abstract:

In order to reveal the influence of the conversion mechanism between garment flat and the pattern, this paper reviewed the processes and methods of making pattern manually according to the garment flat, emphasizing on the recognition methods of the characteristic parameters and the machine learning recognition. Discussions on the pattern conversion technology were carried out based on the garment flat recognition, and its advantages and disadvantages were analyzed. The research shows that parametric and match conversion are the most commonly used conversion methods. Parametric conversion is suitable for clothing with relatively fixed style. The conversion accuracy is high, but different garment flat needs to establish different conversion models. Match conversion can facilitate fast conversion of pattern with high robustness, and permit pattern making rules not to be followed. The disadvantage is that the accuracy is low and large data sets need to be built as training sets at the early stage. The review suggests that in the future, relevant researches should be carried out in three fields, i.e., fining garment flat recognition granularity, garment flat fabric parameter pattern multi-domain matching and componentized pattern generation.

Key words: garment flat, pattern conversion, image recognition, componentized pattern, parametric conversion, match conversion

[1] 许倩, 陈敏之. 基于深度学习的服装丝缕平衡性评价系统[J]. 纺织学报, 2019, 40(10): 191-195.
[2] 黎聪 闫学娜 曾祥忠 梁猛 张莹. 应用一维傅里叶变换的剖幅区自动识别与定位[J]. 纺织学报, 2016, 37(01): 147-151.
[3] 夏帆 刘翔. 基于服装设计要素和法则的智能配搭设计[J]. 纺织学报, 2015, 36(07): 94-99.
[4] 张海波 黄铁军 修毅 赵野军 章江华. 基于神经网络的男西装图像情感语义识别[J]. 纺织学报, 2013, 34(12): 138-0.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
[1] 杨帮华<sup></sup>高晓丁<sup></sup>宋栓军<sup></sup> . 基于虚拟仪器的织机经纱张力测试方法[J]. 纺织学报, 2005, 26(1): 90 -91 .
[2] 成玲;何勇. 印染机械多电机同步控制系统[J]. 纺织学报, 2005, 26(1): 97 -99 .
[3] 高敬玮;陈颖;胡盼盼. 熔喷PBT非织造布血液滤材表面涂层改性[J]. 纺织学报, 2004, 25(04): 9 -11 .
[4] . SGA7101-190型喷气织机生产技术鉴定会[J]. 纺织学报, 1986, 7(04): 56 .
[5] 朱友水;王红卫. 丙纶非织造织物的等离子体金属化处理[J]. 纺织学报, 2005, 26(5): 83 -85 .
[6] . 山西省纺织工程学会动态[J]. 纺织学报, 1981, 2(04): 59 .
[7] 余伟. 铅布织机布边控制装置的设计[J]. 纺织学报, 2004, 25(05): 111 -112 .
[8] 冯毅力;李汝勤. 人体模型的三维数据拾取和服装曲面的生成[J]. 纺织学报, 2004, 25(06): 47 -48 .
[9] 胡心怡;王厉冰. 大豆蛋白纤维织物传热性能研究[J]. 纺织学报, 2003, 24(05): 56 -57 .
[10] 李维贤;赵耀明;师严明. 丝素/甲壳质共溶剂的研究[J]. 纺织学报, 2003, 24(05): 80 -82 .