【作者】 陈龙；

【导师】 陆国栋；

【作者基本信息】 浙江大学 ， 机械设计及理论， 2008， 博士

【摘要】 服装CAD(Garment Computer Aided Design,简称GCAD)领域近年来得到极大关注。传统GCAD基于二维裁片操作,称之为二维GCAD,目前二维GCAD已比较成熟。随着三维设计技术的发展,三维GCAD也成为热点,并已经成为主流技术,但三维GCAD仍存在诸多难点。本文在分析参数化研究历史和存在问题的基础上,首次提出柔性参数化方法的概念。柔性参数化是以自由曲线曲面为基础元素,以满足基础元素之间的约束为目标,以灵活多样的约束柔性求解方法为手段,最终得到尺寸非精确表达模型的一种参数化方法。柔性参数化在三个方面有别于传统的参数化,即元素和模型表示的柔性化、约束的柔性化、求解的柔性化。基于柔性参数化方法的概念,提出服装柔性参数化方法。服装柔性参数化是对服装参数化模型实施多种驱动和联动操作的服装三维设计方法,内容包括模型建构、模型驱动与模型联动。将服装模型分为坯基层、款式层、衣片层、裁片层四个层次,坯基层和裁片层是服装柔性参数化设计的重点。服装模型建构利用特征信息,分层次建构服装模型,得到具有非精确表达的服装几何参数化模型。服装模型驱动利用混合维度信息输入,对服装多层次几何模型进行多因素驱动。服装模型联动基于混合维产品信息,对服装模型进行多层次联动和混合维度联动并最终得到系列化产品。服装模型构建为基础,模型驱动为核心,模型联动为特色,多层次性、多维度性、非精确性三个特性贯穿于服装柔性参数化设计的整个流程。提出基于混合维特征元素构建服装部件模型的技术。参数化人体模型为服装柔性参数化设计提供丰富的混合维特征语义信息。利用人体特征点得到服装特征点,构建服装特征点的拓扑关系,形成粗三角网格模型,通过曲面细分得到服装细分曲面模型。利用人体特征线信息得到服装特征曲线网络,对特征曲线网络施加约束,形成服装特征线框架模型,而后通过三维曲线和混合维曲线插值即可得到服装插值曲面模型。利用人体特征面生成服装特征基曲面,通过对服装基曲面的一系列操作,直接得到服装曲面模型。通过规范化处理,由这三种方法生成的服装模型可以统一为参数化服装模型表示。这三种方法各有优缺点,可以单独使用,也可集成使用。通过对服装曲面模型的后处理和多层次集成建模,从而完成服装模型的构建。提出服装模型的多因素驱动参数化技术。服装模型有尺寸驱动、草图交互驱动、人体驱动三种方式,而人体驱动又可以分为人体整体驱动和人体局部驱动。在尺寸驱动中,尺寸直接驱动相关特征曲线变形;在草图交互驱动中,将二维输入的草图交互信息转化为相关曲线的尺寸值或约束后,才能进一步驱动目标元素变形;在人体驱动方式中,人体整体驱动是利用曲线约束编解码方法,将某一样式的服装特征框架相对于人体模型编码,再相对于其它人体模型解码,然后通过变动人体尺寸来驱动服装模型;人体局部驱动则是利用碰撞检测技术来局部调整服装曲面网格顶点。服装模型驱动分为两个层次:特征框架驱动和曲面驱动。尺寸驱动、草图交互驱动、人体整体驱动都是直接作用于服装特征框架的,特征框架的约束有效性必须得到维护。利用图论方法,构建服装约束图的约束权值矩阵并经过相关处理,获得约束求解路径,进而完成服装模型的柔性驱动。人体局部驱动则利用细分碰撞满足服装交互设计的精度要求,分片分区碰撞检测满足服装设计的速度要求。提出服装模型联动参数化技术。利用展开技术构建服装混合维映射关系,利用裁片映射技术构建裁片和人体映射关系,同时结合服装模型构建流程,从而形成服装模型正向和逆向混合维多层次联动设计技术。基于服装联动设计技术,以裁片放码为例,提出基于混合维度的裁片放码技术。利用裁片边角特性,结合裁片的直角坐标和局部标架两种表达法,构建两种方式的目标函数,通过对这两种目标函数求解,从而达到数值求解的稳定性。以本文的研究成果为技术核心,开发出服装柔性参数化设计原型系统,并集成于三维服装设计系统LooksTailorX。LooksTailorX系统包含人体建模、三维服装设计、裁片展开和处理、裁片缝合和放码四个功能模块。利用该系统,本文给出了人体参数化变形、服装典型部件建模、多因素驱动和多维度参数化联动技术的实例。更多还原

【Abstract】 GCAD (Garment computer aided design) field has attracted great interesting in recent years. Traditional GCAD was based on 2D pattern operations and so called 2D GCAD. 2D GCAD is growing into autumn. Along with 3D design technology development, 3D GCAD turns into hotspot and mainstream, but it faces many difficulties.Based on analysis of the history and problems of parameterization method, this thesis puts forward FPM( Flexible Parameterization Method), a innovative concept which is never advanced by others. FPM is a method which is based on free curve and surface as basic element, aims to satisfy constraints between basic elements, utilizes all kinds of constraint solving methods, then gains a object model expressed by non-accurate sizes. FPM is different with traditional parameterization method in three aspects including expression flexibility, constraint flexibility and solving flexibility.Based on FPM, the thesis presents GFPDM(Garment Flexible Parameterization Design Method). GFPDM is a parameterization method for garment parameterized model by all kinds of drive and united deformation operations. GFPDM includes construction, drive and united deformation for garment model. Garment model can be divided into four layers which include basic geometric layer, style layer, patch layer and pattern layer. The parameterization emphasis is put on the basic geometric layer and pattern layer. Garment model is constructed as a layered model which is expressed by non-accurate dimensions using feature information. Garment model drive changes model with multifactor by usin/g hybrid dimension input information and non-accurate constraint solving. Garment model united deformation gets a series of product based on hybrid dimension product information by deforming multi-layer functional model based on multi layer and hybrid dimension. Modeling as basis, model drive as kernel and model united deformation as trait, features including multilayer, hybrid dimension, non-accurateness run through the whole process of garment design.Advances garment component modeling technology based on hybrid dimension feature elements. Parameterized human body model provides rich semantic information of hybrid dimension feature for garment parameterization. Acquire garment feature points from body feature points and construct topology for the garment feature points, then a coarse triangle mesh will turn up, which will turn into garment mesh through surface subdivision. Utilize human feature line information to get the garment feature curve network, exert constraints on the feature curve network, then the garment feature curve frame will be formed, and then garment surface will be got by surface interpolation through 3D curves and hybrid dimension curves as surface boundary curves. Garment surface model can be produced directly by a series of operation on garment basic surface which is generated from body feature surface. These three kinds of methods each have pluses and minuses. A parameteric garment model can be got from the models modelled by the three ways through standardized treatment. The modeling way based feature lines can get parameterized garment directly, but modeling ways based on feature points and feature surfaces need further treatment.They can be used alone or can be integrated to use. After aftertreatment to the garment surface model and multi-layered integration modeling, the construction of the garment model will finish.Advances multifactor drive parameterization technology for garment parameterization. The garment model has three kinds of drive ways including size drive, sketch drive and human body drive. The human body drive also can be divided into human body’s whole drive and human body’s partial drive. In the way of the size drive, size value urges the relevant feature curves to deform directly. In the way of the sketch drive, after turning the 2D mutual sketch input information into the size value and constraints of the relevant curves, the goal elements can be deformed further. In the way of human body drive, human body’s whole drive utilizes constraints code of curves, which encodes garment feature frame elements referred to a human body and decodes the garment feature elements referred to another human body. Human body’s partial drive utilizes collision detection to adjust local part of garment mesh. To face with model dirve the garment model is also divided into two levels: The feature frame and surface. Size drive, sketch drive and human body’s whole drive all act on garment feature frame directly, so the constraint validity of the garment feature frame must be maintained. Utilize graph theory, construct weight scale matrix for the garment constraint graph, after some relative treatment on the matrix, a solving route will be gained, and flexible drive of the garment model will also finish continuously. Human body’s partial drive utilizes collision detection with surface subdivision to meet the precision request for garment interactive design and technology of dispart and divisional collision detection meets the speed request for garment interactive design.Advances model united deformation technology for garment parameterization. Advances model united deformation technology for garment parameterization. Utilize mesh development technology to construct map relation on hybrid dimension between patch and pattern, use pattern mapping technology to build map relation between pattern and body, then links with garment design modeling process, the design technology named united deformation technology to change garment model forwardly and conversely is formed. Based on this technology, a new grading method based on hybrid dimension is put forward. Making use of information of edge and angle of pattern polygon, two object functions are made out along with two ways including Descartes coordinate and local frame to express polygon. Through solving two ways of object functions, numerical solving robustness can be got.A 3D garment parameterized design software has been developed by exploiting the algorithms presented in this thesis and has been integrated into a garment design system named LookstailorX(LSX). LSX includes four functional modules, which includes mannequin module, garment module, pattern module, sewing and grading module. Using the LSX, many examples including the technologies for body parameterized deformation, modeling, multifactor drive and garment multi dimension united deformation for representative garment component.更多还原