人体冠状面轮廓曲线数学模型构建与量化分析  

Mathematical model construction and quantitative analysis of human coronal profile curve

作　　者：倪世明[1] 白云龙 蒋益群 NI Shiming;BAI Yunlong;JIANG Yiqun(College of Architecture&Art Design,Shaoxing Vocational&Technical College,Shaoxing 312000,China;College of Information Engineering,Shaoxing Vocational&Technical College,Shaoxing 312000,China;College of Engineering,Iowa State University,Iowa 50011,USA)

机构地区：[1]绍兴职业技术学院建筑与设计艺术学院,浙江绍兴312000 [2]绍兴职业技术学院信息工程学院,浙江绍兴312000 [3]艾奥瓦州立大学工程学院,美国艾奥瓦州50011

出　　处：《丝绸》2022年第3期20-27,共8页Journal of Silk

基　　金：浙江省教育厅一般科研项目(Y202045029)。

摘　　要：冠状面轮廓曲线可以表现人体纵向体表形态,是体型研究和原型设计的关键。首先,通过三维人体测量,获取人体点云数据,利用逆向工程技术降维处理,采集冠状面轮廓点云;其次,用最小二乘法、三次样条函数、径向基函数(Radical Basis Function,RBF)神经网络分别进行曲线拟合,比较三个模型的均方根误差RMSE、决定系数R^(2)等参数,得出RBF神经网络拟合效果最优;然后,通过计算拟合曲线的曲率和二阶导数,量化分析冠状面轮廓曲线形态;最后,在大规模人群测量中进行验证,得到3类典型的冠状面曲线。实验结果显示,采用RBF神经网络构建二维曲线表征三维人体特征空间的数学模型拟合精度高、泛化能力强,能有效表征人体纵向体表特征。There are huge individual differences in human body shape,mainly manifested in height(length),fatness and thinness(width and thickness),body surface shape(curve and angle)and so on.Body shape research based on body surface curve can refine the existing body shape classification and is the key to improving clothing fitness.There are three human body sections in sagittal,coronal and transverse directions.Due to the irregular shape of the section curve,it is necessary to build a mathematical model to quantitatively describe the shape characteristics of the curve.The coronal curve shape that contains the body shape information of the neck,shoulder and side of the body is closely related to the clothing structure.However,there are few researches on coronal curves.Despite the common use of methods such as least square method,cubic spline function and RBF neural network in curve fitting,further research is needed for the optimal fitting method of human coronal plane curve.For the purpose of the dimension reduction and optimization of the human body’s high-dimensional point cloud data,a mathematical model with high fitting accuracy and strong generalization ability is established to quantitatively analyze the coronal profile curve,and to better characterize human body surface shape.First,the human body point cloud data are obtained through[TC]^(2) three-dimensional measurement.In order to enhance the universality of the mathematical model of human coronal profile curve,an experimental sample closest to the national standard 160/84A female body size is selected.The original point cloud is simplified,smoothed,projected and symmetrically processed by reverse engineering technology,and the coronal profile point cloud is collected.Secondly,the least square method,cubic spline function and Radical Basis Function(RBF)neural network are used for curve fitting respectively,and the RMSE and determination coefficient R^(2) of the three models are compared.RMSE of RBF neural network is closer to 0 and R^(2) is closer to 1 compared

关 键 词：冠状面轮廓曲线 逆向工程 RBF神经网络 数学模型 量化分析 

分 类 号：TS941.17[轻工技术与工程—服装设计与工程]