Font Size: a A A

Shape Analysis Of Waist-hip Using Feature Distance And Personalized Ffemale Trousers Pattern Making

Posted on:2018-02-07Degree:DoctorType:Dissertation
Country:ChinaCandidate:J F JinFull Text:PDF
GTID:1311330512996131Subject:Textile Science and Engineering
Abstract/Summary:PDF Full Text Request
Personalized garment customization is proposed to meet people's demands on personalization and fitting.How to use 3D body data to generate personalized garment pattern bansed on shape feature has always been a focus issue.Aimed to generate personalized female trousers pattern,how to construct classification indices of three-dimensional shape,automatic recognition of shape model and ‘body-garment' model by using 3D data of waist-hip were discussed in this dissertation,so that the generation of personalized female trousers patter based on waist-hip feature was achicved.In order to get three dimensional morphological characteristics of waist-hip,the type of figure data information of 470 young female in school was obtained using [TC]2 of 3D human body measure system;a total of 10 characteristic cross section curves including thigh circumference,hip line and waist line,were extracted automatically using the bubble sort algorithm to construct feature lines set;mirror image symmetry was done to deal with feature lines,then a point every 15°to cut the close curve into 13 different points,so that a total of 130 points were extracted after centering on the centroid of each characteristic section curve,thus,feature points set was established;according to the Euclidean distance formula,the distance between the feature points and the center of hip line(i.e.the feature distance)was calculated,and the feature distance set was constructed.Aimed to classify and identify the shape type of waist-hip based on typical indices,with the help of MATLAB software,the dimensionality reduction of high-dimensional feature distance set was studied based on manifold learning algorithm,and it was confirmed that the intrinsic dimension value was 5 through residual variance analysis;based on the fast clustering method and the analysis of variance,the optimal dimension reduction method was LE,and the optimal cluster number was 4 types,and then the refined form table was established;by R cluster analysis and correlation analysis,3 typical indicators of Line 1,4 typical indicators of Line 4,3 typical indicators of Line 7 and 4 typical indicators of Line 10 were extracted;the PNN model was constructed to identify the type of waist-hip shape according to the typical indicators by MATLAB software,and the validity of the model was tested by sensitivity,specificity,false positive rate and accuracy.To construct ‘body-garment' model,according to the cluster center,the intermediate point cloud matrix was constructed by using the mean to obtain the 3D coordinates of the intermediate point cloud;using CATIA software,the B spline curves were constructed for each type of intermediate curve model,3D surface model of waist-hip was constructed by lofting technology,and using the NURBS to optimize the surface curves,the aim of constructing human waist-hip model with ‘points-lines-surface' was achieved;by using the statistical method,the regression model between feature distance and basic body sizes(waist circumference,hip circumference and crotch length)was established,the corresponding characteristic distance of trousers was calculated;combining graphics and mathematical knowledge,parameter equations were obtained to established the mathematical model of trousers structure,and the 3D coordinate value was obtained through solving equations to construct type point matrix of trousers,and then 3D surface model of waist-hip of trousers was constructed,the individual differences of trousers model were further verified by the comparison of various types of loose trousers value distribution amount.To generate personalized trousers pattern,according to the basic principle of women trousers structure drawing,the side seam and the waist darts on 3D surface model which had been constructed were determined,the surface flatten was done to generate the initial pattern by rational division of surface region,and then a complete trousers pattern was obtained by modifying the initial model;a comparative analysis of the shape of trousers patterns were done,included back crotch inclination,front crotch inclination,side seam contour,ratio of front waist circumference to back waist circumference,back crotch width and front crotch width,and then verification of the patterns was completed through trying on the sample trousers.Finally,a personalized conversion of 3D model to 2D pattern system framework which including human body data entry module,waist-hip shape data module,waist-hip shape type recognition module,3D model of waist-hip module,personalized 3D model of trousers module and personalized trousers pattern module was designed;the system was completed through connecting each module by VB programming language and reverse engineering software,.
Keywords/Search Tags:Classification Indices of three-dimensional shape, Feature Distance, Manifold Learning, Probabilistic Neural Network(PNN), Female Trousers, personalized pattern
PDF Full Text Request
Related items