Classification Of Young Female's Upper Body Form Based On Space Vector And Pattern Customization

The development of non-contact three-dimensional scanning technology brings a new method for human body measurement,which is used more and more in garment production and scientific research.Although 3D scanner is used to obtain human body data in most of the researches,two-dimensional data(such as circumference,length and width,curvature and so on)are adopted to classify body size,which has many limitations in describing three-dimensional shape.How to meet the fit needs of consumers while reducing garment production costs,is the significance of body classification.However,too many body types will cause too complex of the basic pattern and increase the production costs of enterprises,so we need fewer prototype paper to satisfy more consumers' body characteristics.In this paper,the classification of three-dimensional morphological of young females is carried out from the spatial vector data set.By calculating the vector shape of the center point to the boundary of each cross-section,the classification method of upper body can provide a new mode of thinking.At the same time,we use the random forest algorithm to establish the young female's upper body morphological discrimination model,and finally,generate prototypes of the three sub-categories for the type of 160 / 84 A.The main contents are as follows:1.Data of the human body point cloud and the basic size data were obtained by using the [TC] 2 three-dimensional scanner as subjects of 441 young female non-pregnant females in Jiangsu and Zhejiang provinces;2.The spatial vector data sets were used to classify the female body.The length of the vectors was 240 before principal component analysis,but after that,11 feature length was extracted,which achieve a good dimension reduction effect and explain 95.28%feature;3.Using angle cosine method to determine the optimal number of clusters,the upper body is divided into three categories through the cluster analysis,which are moderate shape,flat shape and round shape,we get the proportion of each subdivided body in the national standard,this three-dimensional shape of different size was described by "national standard body+ subdivided body";4.According to the recognition results of the test samples,which can be seen that the recognition accuracy of the model is very high,and the overall recognition rate is97.59%.The model of the upper body of the young female is identified by the random forest algorithm;5.The corresponding relationship between the prototype model and the upper part of the figure is established for the largest number of 160 / 84 A,and the length of prototype model is estimated by calculating the convex hull length outside the human body.Finally,different sub-body prototype is obtained.