基于极限学习机的乳房形态识别  

Breast shape recognition based on extreme learning machine

作　　者：周捷[1] 王萍 毛倩 王奥斯 ZHOU Jie;WANG Ping;MAO Qian;WANG Aosi(School of Apparel and Art Design, Xi’an Polytechnic University, Xi’an 710048, China;School of Design,University of Leeds LS2 9JT, UK)

机构地区：[1]西安工程大学服装与艺术设计学院,陕西西安710048 [2]利兹大学设计学院,英国利兹LS29JT

出　　处：《西安工程大学学报》2022年第1期17-24,共8页Journal of Xi’an Polytechnic University

基　　金：国家社科基金艺术学项目(20BG134)。

摘　　要：为了提高乳房形态识别精度,采用密度峰值快速聚类(clustering by fast search and find of density peaks,CFSFDP)算法对西部地区108位青年女性的乳房形态特征数据进行聚类分析,再运用极限学习机(extreme learning machine,ELM)算法识别乳房形态,并对比分析了在3种激活函数下,ELM乳房形态识别模型隐含层神经元个数与准确率的关系。结果表明:ELM算法对乳房形态识别准确率较高且用时较短,平均时长为1.28 s。当模型激活函数选择sin且隐含层神经元个数为25时,模型识别乳房形态准确率较好,平均为98.3%。ELM乳房形态识别研究在一定程度上改善了消费者乳房与文胸号型之间的配伍性,为人体形态识别模型参数的选择提供了依据。In view of the low accuracy of breast shape recognition,the approach of CFSFDP was used to cluster the breast shape of 108 young female breast forms in west region,and then the ELM algorithm was used to identify breast shape.The relationship between the number of neurons in the hidden layer and the accuracy of the ELM model under three activation functions was compared and analyzed.The results showed that the breast shape recognition by ELM model has high accuracy and short time,with an average time of 1.28 s.When the activation function of the model was sin and the number of neurons in the hidden layer was set to 25,the average accuracy of breast shape recognition was 98.3%.The research on parameter setting of ELM breast shape recognition model improves the accuracy of breast shape recognition and the compatibility between consumer breast and bra size to a certain extent.It also provides a basis for parameter selection of human body shape recognition research.

关 键 词：乳房形态 乳房识别 极限学习机 文胸 密度峰值快速聚类算法 

分 类 号：TS941.17[轻工技术与工程—服装设计与工程]