三维人体特征点自动提取的研究与应用  

Research and application of automatic extraction of 3D human body feature points

作　　者：姚怡 毛玥[1] 蒙冉菊 金子敏[2] YAO Yi;MAO Yue;MENG Ranju;JIN Zimin(Fashion Institute of Design,Jiaxing Vocational&Technical College,Jiaxing 314036,China;College of Textile Science andEngineering(International Institute of Silk),Zhejiang Sci-Tech University,Hangzhou 310018,China)

机构地区：[1]嘉兴职业技术学院时尚设计学院,浙江嘉兴314036 [2]浙江理工大学纺织科学与工程学院(国际丝绸学院),杭州310018

出　　处：《丝绸》2025年第2期75-81,共7页Journal of Silk

基　　金：教育部人文交流专项研究课题项目(YJYY202308)。

摘　　要：人体特征点准确快速地提取是三维人体尺寸测量、三维人体建模、体型分类等的基础,文章综述了三维人体特征点提取的重要性与困难,以及几何形状分析法、函数法、模板法这三种主要提取方法的原理、特点和应用现状。几何形状分析法计算较为简单,但受体型差异影响较大;函数法方法简单,但拟合精度较低;模板法精度较高但计算量大,且模型库构建成本高。最后,指出了三维人体特征点自动提取的研究应重点突破特征点提取过程中由体型差异、姿态变化等因素带来的难点,开发具备更高鲁棒性、适用性强的自动化算法,以实现更精确、快速的三维人体特征点自动提取。This paper discusses the importance of 3D body scanning technology in the field of apparel and related research,especially how to automatically extract human body feature points from 3D point cloud data to improve the measurement accuracy of human body size data.The traditional manual measurement operation process is cumbersome and time-consuming,while the method of automatically recognizing and extracting the 3D feature points of the human body has obvious advantages in terms of speed and accuracy.However,existing 3D body scanning systems often have feature point recognition errors when dealing with special body sizes,and still need to be adjusted manually.Therefore,the purpose of this paper is to analyze the existing 3D human body feature point extraction methods,explore their difficulties and challenges,and provide technical support for fast and accurate feature point recognition.This paper focuses on three main 3D human feature point extraction methods:geometric shape analysis,function method and template method.The geometric shape analysis method extracts feature points by analyzing the geometric features of the human body feature parts,which is simpler to calculate,but is greatly affected by the differences in human body shapes.The function method constructs the discriminant function by counting the relationship between the feature points,which is simple,but the fitting accuracy is low.The template method matches the point cloud data with predefined human body templates,which is more accurate but computationally intensive and costly to construct the model library.The results of this paper show that the geometric shape analysis method is quite common in practical applications and is suitable for most body types,but the recognition of special body types still needs to be improved.Future research should be devoted to the development of feature point extraction methods that are insensitive to differences in human body size to further improve recognition speed and accuracy.Meanwhile,with the wide application of

关 键 词：三维人体 特征点自动提取 几何形状分析法 函数法 模板法 

分 类 号：TS941.2[轻工技术与工程—服装设计与工程]