Research Of Young Male Body Classiifcation And Recognition Based On Cross Section Shape

In modern garment industry,The GB divides the human shapes according to theThoracolumbar difference, it used to describe the correlation of different size, lack of distinctionbetween the curve shape changes in cross section. The shapes of the curves and the clothingstructure are close relatives, it is necessary to analyze and research the change of the human bodycurve to meet the demand of garment customization and we need to grasp the various parts of thebody accurate. But the study on the cross section shape of curve is very little,it affect thedevelopment and implementation of clothing tailored to a certain.This study conxtructs basic cross-section curve mathematical model on basis of the sectioncurve,and extract the cross section curve morphological parameters to analysis of the humanbody shap, which characterize the human body morphology to provide a new method for thestudy of human body shape. At the same time, the paper constructs a body shape recognitionmodel and human body recognition system on basis of the extreme learning machine,which laythe foundation for the personalized tailored clothing, computer aided design and virtual fitting.This thesis mainly includes the following contents:1.Using the [TC]23D non-contact body measurement instrument subject to measure213male university students,the experimental subjects in the age between18-26years old. Theexperiment collected the three-dimensional point cloud data of the human body, and to extractthe human body dimension data according to the requirements of the project.2.Using imageware to process the point cloud data, and extracted the basic parts sectioncurve characterization that can describe the human body shape. Through the MATLAB softwareachieving cross-section Coordinate transformation、axis adjustment and symmetric treatment.Using least square method to get cross section of curve fitting, and every5degree extractionradius of curvature on the half section.3.On the extraction of radius of curvature were selected.Adopting K-means clusteringalgorithm to dynamic clustering of young male’s body shape on the basis of the cross-sectionfeature points’ radius of curvature and cross-sectional axial ratio. In addition,Throughcorrelation analysis, a regression model is established for the feature points curvature radius canbe characterized by size index that measured conveniently. 4.Using MATLAB R2012b software to build the extreme learning machine (ELM) humanbody recognition model, training and testing of the recognition model. The shoulder of therecognition rate is100%, the chest recognition rate is98%, the waist recognition rate is96%, thehip recognition rate is96%.5.The subject of the development of young male cross-section shape recognition system,system is composed of body size data entry, section curve shape parameter calculation, ELMhuman body recognition, human body curve display. By entering the body size data, calculationof cross-section shape parameters, ELM human body recognition, identification and display ofthe human body.