摘要
首先采用椭圆傅里叶对女性的胸围、腰围和臀围特征断面进行描述,并对椭圆傅里叶描述子进行主成分分析,将各主成分与特征断面的宽度、厚度尺寸进行回归分析,实现由断面宽度、厚度测量尺寸到断面形状的逆向拟合,并根据拟合断面预测围度尺寸。对459个样本的特征断面宽度、厚度和围度拟合误差进行分析,结果显示,拟合误差胸围断面最大,腰围断面最小。最后给出了各个断面的宽度和厚度尺寸输入误差与断面围度尺寸预测误差的关系。研究结果可应用于人体二维摄影测量系统。
Elliptic Fourier method was used to fit female＇s bust, waist and hip slices, and then principal component analysis was applied to the Elliptic Fourier descriptors. Multiple linear regression coefficients of principal components and slice＇s width and depth were calculated to map principal components from slice＇s width and depth. Slices were reconstructed from principal components and girths were derived. Fitting error of slice＇s width, depth and girth of 459 samples were also analyzed. The bust slice has the largest fitting error, and the waist slice, the smallest. Finally, girth fitting errors caused by slice width and depth input error were shown in a regression form. The proposed method could be used in 2-D photographic measuring system.
出处
《纺织学报》
EI
CAS
CSCD
北大核心
2014年第8期76-80,共5页
Journal of Textile Research
关键词
椭圆傅里叶
断面拟合
围度预测
主成分分析
Elliptic Fourier
slice fitting
girth prediction
principal component analysis
