人类学学报 ›› 2021, Vol. 40 ›› Issue (04): 664-678.doi: 10.16359/j.cnki.cn11-1963/q.2020.0011

• 研究论文 • 上一篇    下一篇

基于三维人脸模型及人工测量数据分析现代中国人群面部表型的性别差异

曹家望(), 燕君, 李黎明, 乔辉, 孙畅, 文少卿, 谭婧泽()   

  1. 复旦大学现代人类学教育部重点实验室,复旦大学人类遗传学与人类学系,上海 200438
  • 收稿日期:2019-01-18 修回日期:2019-11-08 出版日期:2021-08-15 发布日期:2021-08-17
  • 通讯作者: 谭婧泽
  • 作者简介:曹家望(1990-),男,学士,学生,人类生物学。Email: caojiawang168@163.com
  • 基金资助:
    国家自然科学基金面上项目(31771325);国家自然科学基金重大研究计划培育项目(91631105);国家自然科学基金重大研究计划集成项目(91731303)

Analysis of sexual dimorphism based on three-dimensional face model and artificial measurement data of maxillofacial morphology in modern Chinese population

CAO Jiawang(), YAN Jun, LI Liming, QIAO Hui, SUN Chang, WEN Shaoqin, TAN Jingze()   

  1. MOE Key Laboratory of Contemporary Anthropology, Department of human genetics and anthropology, Fudan University, Shanghai, 200438
  • Received:2019-01-18 Revised:2019-11-08 Online:2021-08-15 Published:2021-08-17
  • Contact: TAN Jingze

摘要:

传统体质人类学研究中的样本采集方法是手工测量以及肉眼观察,采集过程中会存在较大的主观性,易产生误差,同时在采集大样本时费时费力。近年来,三维摄像系统被越来越多的应用于人脸分析,其优势是快速高效和准确,并且涵盖了传统测量方法无法精确获取的表型特征信息,可以更加细化地分析人脸表型特征和特征之间的相互关系。本研究探索性地基于三维人脸模型图像,利用曲率和邻域信息的点筛查方法定位了面部的一些关键测量点,建立一套简单、有效、准确率较高的三维人脸模型图像关键点定位方法,能较准确地定位98%以上的眼内角点和鼻尖点,并在此基础上确定了鼻根点、鼻下点、左右鼻翼点和鼻翼外侧最低点等其他面部关键点。并且还提出了在三维人脸模型图像基础上量化部分面部观察类表型特征,如鼻侧深度、鼻根高度、面部扁平度等,尝试性地将这些观察类性状转化为能够测量的性状,建立了一套可以计算的描述方法,具有一定的创新性。本研究还结合三维人脸模型提取数据和人工测量数据,探讨了现代中国人群(以江淮汉族为例)面部表型特征的性别差异,发现大多数面部表型特征在现代中国人群中都存在明显的男女性别差异,鼻侧深度、鼻根高度、面部扁平度的性别差异最为显著,其Z检验的p值均小于0.01。结果显示男性的鼻部整体比女性更大更前凸,而女性的面部更为扁平。

关键词: 现代人, 面部, 表型, 性别, 差异

Abstract:

Traditionally, phenotype features collection and measurement analysis in physical anthropology have relied heavily on manual measurements and observational evaluations, for which inter-observer error, and labour-intensity in large-scale sample collection present major challenges. Recently, 3D measuring technology is more and more used as a faster, more efficient and accurate human face phenotyping method which extracts more detailed human face phenotype features and characteristics those could not be captured and measured accurately by the traditional way. Based on 3D human face model, this research introduces an accurate and robust land-marking method by using 3D mesh curvature and vertexes neighborhood feature processing. With this land-marking method, the endocanthions and pronasale are located on the 3D human face mesh with accuracy higher than 98%, additionally, more human face feature points such as nasion, subnasale, left and right alares and the deepest points on the lateral side of alares are located accurately. According to these feature points located on the 3D human face mesh, an efficient quantification analysis system of observational phenotype features is established, features like asal depth, nasion height, facial flatness, etc. are measured and analyzed. In order to analyze sexual dimorphism in Modern Chinese population, phenotype features extracted from 3D human face meshs and manual morphometrics are compared and evaluated. It was found that most of facial phenotype features had gender differences in the Modern Chinese population, and nasal depth, nasion height and facial flatness present the most significant differences, with Value P<0.01 in the Z-Test. The result shows that men’s noses were larger and protruding frontwards than women’s, while women’s faces are flatter accordingly.

Key words: Modern population, Maxillofacial, Morphology, Sex, Dimorphism

中图分类号: