摘要
使用非接触式三维人体测量系统对中国北方青年女性人体进行测量,测量数据包括肩宽、胸围、臀围等部位尺寸,这些尺寸作为特征值构成了人体体型特征向量。不均匀体型的比例愈来愈高,因此更进一步细化对人体体型的分类识别变得十分重要。考察了某些特征尺寸的概率分布,然后利用聚类分析将人体体型分为偏瘦、正常、偏胖3类,最后提出使用马氏距离的方法分析体型匀称问题。实验结果表明,所使用的方法可有效实现人体体型的细分识别。
A non-contact 3D body measurement system was used to perform measurements of young females from north China,and the measured data included the size of shoulder,bust,and hip width etc.These dimensions are used as eigenvalues,constitute feature vectors of bodily form.In recent years,excessive uncoordinated sizes of partial body appeared in Chinese people.So it′s important to detail the human body segment recognition.First,the probability distributions of certain feature size were investigated,and then the human bodily forms were divided into three types of underweight,normal,overweight,using clustering analysis of K-mean.Finally,it is suggested to use Mahalanobis Distance to analyze the problem of body symmetry.The experimental results demonstrate that this method is effective for detailing the human body segment recognition.
出处
《纺织学报》
EI
CAS
CSCD
北大核心
2011年第2期107-111,120,共6页
Journal of Textile Research
关键词
三维人体测量系统
K-mean聚类分析
马氏距离
人体体型细分识别
3D body measurement system
clustering analysis of K-mean
Mahalanobis Distance
recognition of human body segments
