Font Size: a A A

Research On Multi-level Linkage Parametric Modeling Method Of 3D Clothing

Posted on:2019-08-30Degree:MasterType:Thesis
Country:ChinaCandidate:M HuangFull Text:PDF
GTID:2438330566483720Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
At present,the research of 3D garment modeling has become a hot topic at home and abroad.With the rapid development of computer network technology and3 D display equipment,there are more needs in the application of e-commerce,3D virtual try on and computer animation entertainment.There are interactive 3D garment modeling methods,data driven 3D modeling methods,3D garment reverse process modeling methods,high level analysis and processing methods.Because of the different modeling principles and methods,they have their own shortcomings.In order to improve the 3D garment modeling accuracy and efficiency,we propose a 3D garment multi-level linkage study on parametric modeling including semi supervised learning based 3D garment segmentation method,the geometric constraint of the linkage of 3D garment reconstruction,3D garment modeling and simulation system of three parts.In this paper,a 3D garment segmentation method based on semi supervised learning is first proposed.Firstly,the objective function based on conditional random field model(CRF)is defined for input labeled and unlabeled garment model datasets.Then,the JointBoost classifier is modified to train the CRF model.Finally,the relevant parameters are learned through iterative optimization,and the final 3D garment segmentation results are obtained by using belief propagation(BP).This method can get a better result of 3D garment segmentation and has a certain robustness.Finally,based on the three-dimensional garment components established,the geometric constraint conditions and linkage attributes established between the components are used to achieve the fusion of 3D garment components.In the process of fusion processing,mathematical models such as segmentation,fusion and conformal transformation of 3D geometric shapes should be established to establish a smooth three-dimensional garment because of the adaptability of geometric controls.In view of the complexity of 3D garment models used in various applications,the efficiency of 3D garment modeling is higher.In this paper,a 3D garment reconstruction method based on geometric linkage constraints is proposed.First,through the normalization of the basic control of clothing,the topological geometric mapping relationship between the two-dimensional garment template and the 3D garment model is established,and then a three-dimensional component parameterized model based on the linkage attributes of the components isestablished.The model is a structural graph with hierarchical characteristic relations,and its directional relation represents the linkage constraint condition for 3D geometric reconstruction.A 3D garment modeling and simulation system has been developed,and a prototype system is developed from two-dimensional sample to 3D garment.It includes four modules: 2-D and 3D template conversion module,three dimensional window management,3D modeling of clothing,and 3D simulation clothing demonstration.Three-dimensional clothing can be built simply and effectively.
Keywords/Search Tags:3D garment modeling, Semi-supervised learning, Garment segmentation, Mesh merging, Parametric Modeling
PDF Full Text Request
Related items