纺织学报 ›› 2020, Vol. 41 ›› Issue (12): 111-117.doi: 10.13475/j.fzxb.20200502607

• 服装工程 • 上一篇    下一篇

基于二维图像的青年女性颈肩部形态自动识别

王婷1, 顾冰菲1,2,3()   

  1. 1.浙江理工大学 服装学院, 浙江 杭州 310018
    2.浙江省服装工程技术研究中心, 浙江 杭州 310018
    3.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, 浙江 杭州 310018
  • 收稿日期:2020-05-13 修回日期:2020-08-31 出版日期:2020-12-15 发布日期:2020-12-23
  • 通讯作者: 顾冰菲
  • 作者简介:王婷(1995—),女,硕士生。主要研究方向为数字化服装技术。
  • 基金资助:
    国家自然科学基金项目(61702461);国家自然科学基金项目(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)

Automatic identification of young women's neck-shoulder shapes based on images

WANG Ting1, GU Bingfei1,2,3()   

  1. 1. School of Fashion Design & Engineering, Zhejiang Sci-Tech University, Hangzhou, Zhejiang 310018, China
    2. Clothing Engineering Research Center of Zhejiang Province, Hangzhou, Zhejiang 310018, China
    3. Key Laboratory of Silk Culture Inheriting and Products Design Digital Technology, Ministry of Culture and Tourism, Hangzhou, Zhejiang 310018, China
  • Received:2020-05-13 Revised:2020-08-31 Online:2020-12-15 Published:2020-12-23
  • Contact: GU Bingfei

摘要:

为实现青年女性颈肩部形态的自动识别,首先基于202名在校青年女性的三维点云数据,测量了15个颈肩部形态相关参数,通过分析确定出离散程度较大的形态参数,包括肩斜角、背入角、肩矢额径比和腋下矢额径比;然后结合这4个重要体型参数,对青年女性颈肩部形态进行细分并建立各类体型的分类规则;最后基于青年女性正面与侧面二维照片,通过提取人体轮廓和识别特征点获得颈肩部体型分类所需参数,根据体型分类规则实现颈肩部形态的自动识别。结果表明:青年女性颈肩部形态可分为4类,即圆宽肩体、扁窄肩体、圆落肩体、驼背扁肩体,分别占样本总数的25.53%、23.94%、25.59%和23.94%;通过对40名测试样本进行基于正、侧面二维照片的颈肩部形态自动识别验证,准确率达到90%,说明基于本文方法构建的颈肩部体型自动识别系统是有效的。

关键词: 颈肩部形态, 体型分类, 图像, 尺寸提取, 自动识别

Abstract:

In order to facilitate the automatic identification of young women's neck-shoulder shapes, 15 neck-shoulder shape parameters of 202 young women were measured in the form of three-dimensional point cloud data, and the parameters with a large degree of dispersion were determined through analysis, including the shoulder angle, back angle, shoulder depth/width ratio and armpit depth/width ratio. Combined with these four important body parameters, the neck-shoulder shape of young women was classified following the established classification rules. Based on the front and side images, the parameters required for neck-shoulder shape classification were obtained by extracting the human body contour and identifying the feature points, and the automatic identification of the neck-shoulder shape was achieved according to the body type classification rules. The results show that young women's neck-shoulder shape is divided into four categories, namely round wide shoulder, flat narrow shoulder, round drop shoulder, hunchback flat shoulder, accounting for 25.53%, 23.94%, 25.59% and 23.94%, respectively, of the total sample. The identification of the neck-shoulder shape based on the front and side images of 40 test samples is verified, and the accuracy ratio reaches 90%, indicating that the neck-shoulder shape automatic identification system constructed using this method is effective.

Key words: neck-shoulder shape, body classification, image, size extraction, automatic identification

中图分类号: 

  • TS941.17

表1

颈肩部形态参数测量和计算"

序号 指标 测量与计算方法 序号 指标 测量与计算方法
1 身高(H) 头顶点至地面的垂直距离 8 颈高比(RhNP) 颈点高(HNP)/身高(H)
2 颈点高(HNP) 侧面头部与颈部交接点至地面的垂直距离 9 侧颈高比(RhSNP) 侧颈点高(HSNP)/身高(H)
3 侧颈点高(HSNP) 正面颈与肩交接点至地面的垂直距离 10 肩高比(RhSP) 肩点高(HSP)/身高(H)
4 肩点高(HSP) 肩端点至地面的垂直距离 11 腋高比(RhAP) 腋点高(HAP)/身高(H)
5 前腋点高(HAP) 手臂与胸部交接点至地面的垂直距离 12 颈矢额径比(RNP) 颈厚(DNP)/颈宽(WNP)
6 肩斜角(AST) 侧颈点和肩端点连线与水平面形成的夹角 13 侧颈矢额径比(RSNP) 侧颈厚(DSNP)/侧颈宽(WSNP)
7 背入角(ADE) 侧面背部最突出的点和颈点的连线
与垂直面形成的夹角
14 肩矢额径比(RSP) 肩厚(DSP)/肩宽(WSP)
15 腋下矢额径比(RAP) 腋下厚(DAP)/腋下宽(WAP)

图1

颈肩部形态参数测量方法"

图2

正态P-P图"

表2

相关变量描述统计分析表"

变量描述 身高/
cm
颈点
高/cm
侧颈点
高/cm
肩点高/
cm
前腋点
高/cm
肩斜角/
(°)
背入角/
(°)
颈矢额
径比
侧颈
矢额径比
肩矢额
径比
腋下
矢额径比
最小值 151.1 128.5 125.6 122.9 110.0 15.9 11.1 0.80 0.61 0.31 0.51
最大值 175.0 150.5 148.5 149.2 136.3 32.0 28.9 1.20 0.99 0.51 0.82
平均值 161.5 137.8 135.4 132.7 122.2 25.1 17.7 0.99 0.78 0.41 0.66
标准差 4.6 4.3 4.1 5.0 4.5 3.1 3.5 0.08 0.06 0.04 0.07
变异系数/% 2.848 3.120 3.028 3.768 3.682 12.351 20.468 8.081 7.692 9.756 10.606

表3

最终聚类中心"

体型
分类
肩斜角/
(°)
背入角/
(°)
肩矢额
径比
腋下矢
额径比
人数
a 24.0 16.5 0.42 0.72 48
b 23.2 15.7 0.37 0.61 45
c 28.7 16.6 0.45 0.68 50
d 24.4 21.9 0.41 0.62 45

图3

4类体型横截面形态"

表4

4类颈肩部体型分类规则"

体型
分类
肩斜
角/(°)
背入
角/(°)
肩矢额
径比
腋下矢
额径比
a 19~27
(不含27)
13~19
(不含19)
0.38~0.46 0.66~0.82
b 16~26 11~18 0.31~0.41 0.51~0.66
(不含0.66)
c 27~32 13~21 0.40~0.51 0.58~0.77
d 19~27
(不含27)
19~29 0.36~0.46 0.54~0.69

表5

判别准确性分析"

体型
分类
预测的群成员分类 总计 判别准
确率/%
a b c d
a 46 0 2 0 48 95.8
b 1 44 0 0 45 97.8
c 3 0 46 1 50 92.0
d 0 1 0 44 45 97.8

图4

校准示意图"

图5

拍摄姿势示意图"

图6

正侧面图像处理过程"

图7

颈点确定示意图"

图8

肩斜角计算示意图"

图9

本文分类规则对颈肩部识别结果"

表6

特征形态参数提取误差分析"

变量 类型 均值 标准差 相关系数 平均绝对误差 误差范围
肩斜角 提取值 25.296° 2.405 0° 0.955 0.63° -1.3°~1.4°
测量值 25.336° 2.451 6°
背入角 提取值 18.201° 2.167 2° 0.891 0.96° -1.5°~1.7°
测量值 18.130° 2.422 6°
肩矢额径比 提取值 0.406 2 0.455 3 0.927 0.01 -0.02~0.02
测量值 0.410 9 0.255 5
腋下矢额径比 提取值 0.642 0 0.563 3 0.776 0.03 -0.07~0.06
测量值 0.630 0 0.555 6

表7

特征形态参数配对样本T检验结果"

变量 均值 标准差 标准误差 T 显著性
肩斜角 -0.040 0.727 0.115 -0.352 0.727
背入角 0.705 1.102 0.174 0.405 0.688
肩矢额径比 -0.005 0.024 0.004 -1.249 0.219
腋下矢额径比 0.011 0.047 0.008 1.438 0.158
[1] YAN J, KUZMICHEV V E. A virtual e-bespoke men's shirt based on new body measurements and method of pattern drafting[J]. Textile Research Journal, 2020,90(19/20):2223-2244.
doi: 10.1177/0040517520913347
[2] 龚昱. 基于形状上下文的体型分类和自动识别研究[D]. 上海: 东华大学, 2017: 1-5.
GONG Yu. Research of body shape classification and recognition based on shape context[D]. Shanghai: Donghua University, 2017: 1-5.
[3] 李小辉, 张思严, 赵楚瞳, 等. 衣身肩斜角与人体体型特征的关系[J]. 纺织学报, 2018,39(11):111-115.
LI Xiaohui, ZHANG Siyan, ZHAO Chutong, et al. Relationship between garment shoulder angle and characteristics of human body[J]. Journal of Textile Research, 2018,39(11):111-115.
[4] 蒋黎. 基于人体肩部形态的女衬衣肩线结构设计研究[D]. 武汉: 武汉纺织大学, 2018: 29-44.
JIANG Li. Research on pattern design of blouse shoulder line based on female body shape[D]. Wuhan: Wuhan Textile University, 2018: 29-44.
[5] 贺莉文. 基于肩截面曲线的弓背型青年男体细分与识别模型[D]. 苏州: 苏州大学, 2018: 27-32.
HE Liwen. Classification and recognition model of the hunchback young male body based on shoulder section curve[D]. Suzhou: Soochow University, 2018: 27-32.
[6] LEONG I F, FANG J J, TSAI M J. A feature-based anthropometry for garment industry[J]. International Journal of Clothing Science & Technology, 2013,25(1):6-23.
[7] UHM T, PARK H, PARK J I. Fully vision-based automatic human body measurement system for apparel application[J]. Measurement, 2015,61:169-179.
doi: 10.1016/j.measurement.2014.10.044
[8] WANG L, WAN T R, TANG W, et al. An efficient human body contour extraction method for mobile apps[C]// International Conference on Technologies for E-Learning and Digital Entertainment. Cham: Springer, 2017: 173-181.
[9] PRADHAN L, SONG G, ZHANG C, et al. Feature extraction from 2D images for body composition ana-lysis[C]// IEEE International Symposium on Multimedia. Miami: IEEE, 2015: 45-52.
[10] JIANG L, YAO J, LI B, et al. Automatic body feature extraction from front and side images[J]. Journal of Software Engineering & Applications, 2013,5(12):94-100.
[11] 马黎. 基于图像的非接触式人体测量算法的研究与实现[D]. 成都: 电子科技大学, 2017: 36-47.
MA Li. Research and implementation of non-contacting anthropometric algorithm based on image[D]. Chengdu: University of Electronic Science and Technology, 2017: 36-47.
[12] 胡新荣, 刘嘉文, 刘军平, 等. 基于ABSS的着装人体多特征点提取与尺寸测量[J/OL]. 计算机工程: 1-10 [2020-01-10]. http://kns.cnki.net/kcms/detail/31.1289.tp.20200110.1329.004.html.
HU Xinrong, LIU Jiawen, LIU Junping, et al. Multi-features extraction and dimension measurement for dressed human bodies via adaptive body structure segmentation[J/OL]. Computer Engineering: 1-10 [2020-01-10]. http://kns.cnki.net/kcms/detail/31.1289.tp.20200110.1329.004.html.
[13] SENANAYAKE M, RAHEJA A, ZHANG Y H. Automated human body measurement extraction[J]. International Journal of Clothing Science & Technology, 2018,30(2):175-194.
[14] 许家岩, 匡才远. 体表特征参数对远程人体测量精度的影响[J]. 丝绸, 2019,56(8):36-45.
XU Jiayan, KUANG Caiyuan. Influence of characteristic parameters of body surface on long-range body measurement precision[J]. Journal of Silk, 2019,56(8):36-45.
[15] 雷启然, 刘琴, 尚笑梅. 人体肩斜度不同测量方法的优化分析[J]. 北京服装学院学报(自然科学版), 2019,39(3):16-23.
LEI Qiran, LIU Qin, SHANG Xiaomei. Optimization analysis of different measurement methods of shoulder inclination[J]. Journal of Beijing Institute of Fashion Technology(Natural Science Edition), 2019,39(3):16-23.
[16] 侯梦婕. 基于三维人体扫描枕型的结构研究[D]. 西安: 西安工程大学, 2019: 15-31.
HOU Mengjie. Research on pillow structure based on three-dimensional scanning of human body[D]. Xi'an: Xi'an Polytechnic University, 2019: 15-31.
[17] 金娟凤, 庞程方, 陈伟杰, 等. 青年男性肩点横截面曲线及其体型细分[J]. 纺织学报, 2016,37(8):100-106.
JIN Juanfeng, PANG Chengfang, CHEN Weijie, et al. Study on subdivision of young male's shoulder shapes and cross-section curve[J]. Journal of Textile Research, 2016,37(8):100-106.
[18] GU B, LIU G, XU B. Girth prediction of young female body using orthogonal silhouettes[J]. The Journal of The Textile Institute, 2017,108(1):140-146.
doi: 10.1080/00405000.2016.1160756
[1] 王婷, 顾冰菲. 基于图像的人体颈肩部三维模型构建[J]. 纺织学报, 2021, 42(01): 125-132.
[2] 裘柯槟, 陈维国, 周华, 应双双. 成像技术在纺织品颜色测量中的应用进展[J]. 纺织学报, 2020, 41(09): 155-161.
[3] 李涛, 杜磊, 黄振华, 蒋玉萍, 邹奉元. 服装款式图识别与样板转换技术研究进展[J]. 纺织学报, 2020, 41(08): 145-151.
[4] 邵金鑫, 张宝昌, 曹继鹏. 基于图像处理与深度学习方法的棉纤维梳理过程纤维检测识别技术[J]. 纺织学报, 2020, 41(07): 40-46.
[5] 刘婷婷, 徐红, 梅馨元, 刘一心, 肖爱民. 基于XGBoost算法对新疆女性臀部体型判别及原型修正[J]. 纺织学报, 2020, 41(07): 147-153.
[6] 余佳佳, 李健. 中国东部地区青年女性人体体型分类[J]. 纺织学报, 2020, 41(05): 134-139.
[7] 周捷, 毛倩. 基于改进快速搜索和发现密度峰值算法的男童体型分类及判别[J]. 纺织学报, 2020, 41(05): 146-152.
[8] 王泽霞, 陈革, 陈振中. 基于改进卷积神经网络的化纤丝饼表面缺陷识别[J]. 纺织学报, 2020, 41(04): 39-44.
[9] 路浩, 陈原. 基于机器视觉的碳纤维预浸料表面缺陷检测方法[J]. 纺织学报, 2020, 41(04): 51-57.
[10] 潘博, 钟跃崎. 基于二维图像的三维服装重建[J]. 纺织学报, 2020, 41(04): 123-128.
[11] 张爱丹, 周赳. 聚集态网点结构提花织物的灰度仿真特性[J]. 纺织学报, 2020, 41(03): 62-67.
[12] 景军锋, 张君扬, 张缓缓, 苏泽斌. 梯度空间下的丝饼表面缺陷检测[J]. 纺织学报, 2020, 41(02): 44-51.
[13] 张铮烨, 辛斌杰, 邓娜, 陈阳, 邢文宇. 基于边界跟踪测量麻纤维横截面参数的算法研究与应用[J]. 纺织学报, 2020, 41(02): 39-43.
[14] 孙洁, 丁笑君, 杜磊, 李秦曼, 邹奉元. 基于卷积神经网络的织物图像特征提取与检索研究进展[J]. 纺织学报, 2019, 40(12): 146-151.
[15] 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.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!