Research On Young Male's Body Shape Classification And Measurement Based On Body Features

The development of Internet garment sales is rapid, but the shortcoming of garment size series and difficulty of getting body data impact its further development. The garment size series of national standard (GB), which based on difference between chest and waist girth, failed to reflect the detailed characteristics of the human body shape. Moreover, conventional method and 3D measurement method have limitation in the Apparel E-Commerce.Therefore, it's significant to exploring reasonable method of classification and measurement.In this study, it foused on young male's body shape characters and classification by body digital images taken by camera. Image procession technique was used for measuring body dimensions. This method was a non-contact measurement method, which was expected to meet the flexible requirement of modern garments production and keep pace with the development of Internet fashion marketing.Hereby, the following five aspects were included in this study:1. In this study, young male college students in Shanghai, Jiangsu Province and Zhejiang Province were subjects. Measuring image, conventional method and 3D non-contact method were used to measure anthropometric dimensions separately. 333 subjects'data were obtained and their body shape characters were concluded.2. First classification was performed to divide subjects'body shape by three determining feature angles that were selected by correlation analysis. Second classification divided the subjects into 28 groups. Compared two ways, classification combined angle with difference between chest and waist girth, and national standard classification which based on difference between chest and waist girth, found that the fore one was more reasonable.3. The silhouettes of human body images were extracted,which based on image technology; the feature points on the silhouettes were identified to obtain girth data. The result of consistency test between extracted value and manual measured value demonstrated classification put forward in this study was feasible.4. Using Back-Propagation Neural Network and regression analysis, the model was set up to predict neck circumference, chest circumference, waist circumference, belt circumference, hip circumferenc, etc., the feasible formulas were selected by checking.5. Young male anthropometric dimensions database was built by C# programming language. The functions of database include data searching, type identifying, anthropometric dimension forecasting etc.