A quantitative classification of body shape by 3-D measurement data

由于肩部支撑着服装，其贴合性影响服装的舒适性，因此本研究的目的在于显示年轻女性的肩部形态，建立肩部形态类型分类的方法，并通过三维测量数据对肩部形态进行分类。首先，为了确定最佳摄影条件和用于本研究的非接触式三维数字化仪的精度，进行了三个实验。结果表明，摄像机的最佳水平和角度是确定的，三维数据的精度是好的，但需要注意人体的晃动。接着，于2004年11月至12月间，以3D数位机拍摄63位女性学生志愿者的照片。分析三维图像和从三维坐标值计算的19个项目，以检查肩的平均形状和个体差异。然后我们应用了主要的比较方法， 关于我们 对11个项目进行因子分析，提取表征肩形的因子，并尝试用聚类分析的因子得分对肩形进行分类，结果表明：3D图片显示的肩形多样，肩颈部位的大小个体差异较大。肩前曲与肩斜、颈基深与颈基宽的相关性很低，是独立的。作为主成分分析的结果，我们提取了三个主成分：第一个成分，表示肩的大小和肩前曲的程度，第二个成分，表示肩斜度，颈底斜度，前胸圆度，和第三个成分，表示前胸宽和颈底宽。然后运用聚类分析的方法将肩关节形态分为4类。只有两个成员属于第四组，所以我们试图解释其余三组。第一组肩型中等，第二组肩型方肩、颈底斜度小、胸型丰满，第三组肩宽肩深小、后宽相对窄。少

This study aimed to show the young women's shoulder shape, to establish the method of shoulder shape type classification, and to classify the shoulder shape by 3-D measurement data, because the shoulder part supports clothing and the fitting to this part influences the comfortableness of clothing.First, in order to confirm the optimum photography conditions and the precision of a non-contact 3-D digitizer used for this study, three experiments were conducted. As the results, the optimum level and angle of camera were clarified, and it was confirmed that the precision of 3-D data was good, however, it was indicated that the cautions to the sway of human body are required. Next, sixty-three female student volunteers were photographed by 3-D digitizer from November to December in 2004. The three-dimensional pictures and the 19 items computed from the 3-D coordinate values were analyzed to examine the average shoulder shape and the individual differences. Then we applied the principal comp … More onent analysis to the 11 items to extract factors which express the shoulder shapes, and tried to classify the shoulder shapes, using the component scores by cluster analysis.The main results are as follow ; their shoulder shapes by 3-D pictures are various and the sizes on shoulder and neck parts have large individual differences. The correlations between shoulder curvature forward and shoulder slope and between neck base depth and width are very low, therefore, they are independent. As the results of principal component analysis, we extracted three principal components ; the first component which express the shoulder size and the degree of shoulder curvature forward, the second component which express the shoulder slope, the neck-base slope, and the roundness of front chest, and the third component which express the front chest width and the neck-base width. Then using cluster analysis of component score, we classified their shoulder shapes to 4 groups. Only two members belong to the fourth group, so we tried to interpreter about the rest three groups. The first group has the average shoulder type, the second group has the square shoulder, the small neck-base slope, and the full bust, and the third group has the small shoulder width and depth and the narrow back width relatively. Less