基于连衣裙结构特征匹配的款式识别技术

doi:10.13475/j.fzxb.20190902306

纺织学报 ›› 2020, Vol. 41 ›› Issue (07): 141-146.doi: 10.13475/j.fzxb.20190902306

基于连衣裙结构特征匹配的款式识别技术

夏明¹^,²^,³(), 宋婧¹^,²^,³, 姜朝阳¹, 马延彬¹

1.东华大学服装与艺术设计学院, 上海 200051
2.同济大学上海国际设计创新研究院, 上海 200092
3.东华大学现代服装设计与技术教育部重点实验室, 上海 200051

收稿日期:2019-09-06 修回日期:2020-04-16 出版日期:2020-07-15 发布日期:2020-07-23
作者简介:夏明(1981—),男,副教授,博士。主要研究方向为服装数字化技术。E-mail: xiaming@dhu.edu.cn。
基金资助:
中央高校基本科研业务费专项资金资助项目(2282020G-08);中央高校基本科研业务费专项资金资助项目(2232019G-08);上海市浦江人才计划资助项目(18PJC002)

Style recognition technique based on feature matching in dress construction

XIA Ming¹^,²^,³(), SONG Jing¹^,²^,³, JIANG Zhaoyang¹, MA Yanbin¹

1. College of Fashion and Art Design, Donghua University, Shanghai 200051, China
2. Shanghai Institute of Design and Innovation, Tongji University, Shanghai 200092, China
3. Key Laboratory of Clothing Design & Technology, Ministry of Education, Donghua University, Shanghai 200051, China;

Received:2019-09-06 Revised:2020-04-16 Online:2020-07-15 Published:2020-07-23

摘要/Abstract

摘要：

为便于人们快速检索相似款式的服装,以连衣裙款式关键点轮廓为基础,对轮廓的不同描述和相似性度量计算的检索方法进行了研究。将形状上下文、传统欧式距离和基于系数分配的欧式距离3种算法计算的相似度结果与人工测评的相似度测评表做对比,确定最优算法。结果表明:形状上下文匹配方法的效果最佳,其中准确度和精确度分别为82.91%和77%,可作为轮廓相似性度量算法;基于系数分配的欧式距离与传统欧式距离的匹配效果相比,准确度和精确度都有所提升,分别提高6.45%和0.44%;根据计算初步推断,在轮廓特征描述子中按照相似度影响程度大小排序为:角度余弦值、宽度、高度。

关键词: 服装轮廓, 连衣裙, 款式识别, 形状上下文, 欧式距离

Abstract:

In order to facilitate rapid retrieval of similar styles of clothing, this paper reports a study on different descriptions of dress contour and on the retrieval methods for measuring and calculating dress similarity, based mainly on the key contour point of dresses. The similarity results were calculated by the three algorithms, i.e., the shape context, traditional Euclidean distance and Euclidean distance based on coefficient assignment, and the results were compared to identify the optimal algorithm. The results show that the shape context matching method has the best effect, and the accuracy and precision are 82.91% and 77%, respectively, which can be used as the contour similarity measure algorithm. The accuracy and precision of the Euclidean distance based on the coefficient allocation are improved by 6.45% and 0.44% respectively. Based on calculation, it is preliminarily concluded that the degree of similarity influence, represented by a contour feature descriptor, is ranked in the order of angle cosine value, width and height.

Key words: clothing contours, dress, style recognition, shape context, Euclidean distance

中图分类号:

TS941.17

夏明, 宋婧, 姜朝阳, 马延彬. 基于连衣裙结构特征匹配的款式识别技术[J]. 纺织学报, 2020, 41(07): 141-146.

XIA Ming, SONG Jing, JIANG Zhaoyang, MA Yanbin. Style recognition technique based on feature matching in dress construction[J]. Journal of Textile Research, 2020, 41(07): 141-146.

图/表 6

图1

图2

图3

图4

图5

图6

参考文献 10

[1]	司文. 服装廓形与结构设计研究[D]. 大连:大连工业大学, 2016: 22-41.
	SI Wen. Clothing profile and structure design research[D]. Dalian:Dalian Polytechnic University, 2016: 22-41.
[2]	黄英. X形连衣裙分割线的设计研究[D]. 武汉:武汉纺织大学, 2013: 24-47.
	HUANG Ying. Design and research of X line of contour in the shape of the dress[D]. Wuhan:Wuhan Textile University, 2013: 24-47.
[3]	倪文杨. 连衣裙产品开发中廓形结构的规划性设计研究[D]. 杭州:浙江理工大学, 2017: 8.
	NI Wenyang. The research on the planning and designing of the silhouette structure in the development of the dress[D]. Hangzhou:Zhejiang Sci-Tech University, 2017: 8.
[4]	WANG Z, NEWTON E, NG R. Ease distribution in relation to the X-line style jacket: part 1: development of a mathematical model[J]. Journal of The Textile Institute, 2006,97(3):247-256.
[5]	HOU B, GUAN H, JIANG J G, et al. SAR image despeckling based on improved directionlet domain gaussian mixtures model[C] //2011 IEEE International Geoscience and Remote Sensing Symposium. Canada: IGARSS, 2011: 3795-3798.
[6]	陶晨, 段亚峰, 印梅芬. 服装廓形的识别与量化[J]. 纺织学报, 2015,36(5):79-82.
	TAO Chen, DUAN Yafeng, YIN Meifen. Identification and quantification of apparel silhouette[J]. Journal of Textile Research, 2015,36(5):79-82.
[7]	MA S, DU T. Improved adaboost face detection[C] //2010 International Conference on Measuring Technology and Mechatronics Automation (ICMTMA 2010). Changsha: IEEE, 2010: 434-437.
[8]	陈嫒嫒, 李来, 刘光灿, 等. 基于关键点的服装检索[J]. 计算机应用, 2017,37(11):3249-3255.
	CHEN Aiai, LI Lai, LIU Guangcan, et al. Clothing retrieval based on landmarks[J]. Journal of Computer Applications, 2017,37(11):3249-3255.
[9]	屠青青. 基于关键点服装款式识别的智能时尚推荐系统研究[D]. 成都:电子科技大学, 2014: 27-32.
	TU Qingqing. Research of key point based style recognition and intelligent fashion clothing recommendation system[D]. Chengdu:University of Electronic Science and Technology of China, 2014: 27-32.
[10]	徐海博. 基于上下文相关的图像检索方法研究[D]. 大连:大连理工大学, 2014: 16-22.
	XU Haibo. Research on context-sensitive image retrieval[D]. Dalian:Dalian University of Technology, 2014: 16-22.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

基于连衣裙结构特征匹配的款式识别技术

Style recognition technique based on feature matching in dress construction

RichHTML

PDF (PC)

摘要/Abstract

引用本文

使用本文

图/表 6

参考文献 10

相关文章 4

Metrics

本文评价

推荐阅读 0

[1]	陈丽丽, 刘成霞. 面料及运动速度对连衣裙三围形态的影响[J]. 纺织学报, 2020, 41(04): 135-141.
[2]	李东万贤福汪军. 采用傅里叶描述子和支持向量机的服装款式识别方法[J]. 纺织学报, 2017, 38(05): 122-127.
[3]	何甜吴志明. 连衣裙用多梳拉舍尔定位蕾丝面料的纹样设计[J]. 纺织学报, 2016, 37(08): 114-118.
[4]	丁锡强;刘国伟. 服装斜裁技术的应用[J]. 纺织学报, 2006, 27(8): 98-100.