面向个性化定制产品的人像三维建模技术及其应用研究
发布时间：2018-11-26 08:54
【摘要】：制造业是国民经济的支柱产业。制造业发展的一个重要趋势就是不断开发个性化定制产品,以满足快速发展的市场需求。当前,一类与人像有关的个性化定制产品随之产生,正逐渐形成一个潜在的巨大新兴产业。 与人像有关的个性化定制产品的关键问题之一是人像三维建模。如何获取逼真的三维人像模型是计算机辅助设计、计算机图形学以及先进制造技术等领域中最具挑战性的问题之一。人像三维建模最直接的方法是利用三维扫描设备获取人像的三维点云,该方法虽然精度较高,但存在设备造价高、建模对象扫描不便、头发对光的反射而难以扫描等问题。目前最主要的手段是利用二维图像重构人像三维模型,已广泛用于人物动画、影视广告、游戏制作等虚拟场景中,所建立的模型主要是一种满足视觉效果的纹理模型。但是这种人像三维模型要应用到个性化定制人像产品开发中,仍存在如下几个难点问题:1)所建立的模型不能直接用于产品加工制造;2)所建立的模型在特征点处有着较高的重构精度,但在非特征区域和特征线处误差较大,模型相似度不能满足产品的开发;3)缺少一种人像产品的顾客满意度评价方法。 针对上述问题,本文主要研究如何利用二维图像建立面向加工制造的相似度高的人像三维模型及其在个性化定制人像产品开发中的应用,为形成一种新的产业提供理论方法和技术支持。研究内容主要包括以下几个方面: 首先,论述了个性化定制人像产品,建立了面向个性化定制产品的人像三维建模技术体系总体框架,包括建模关键技术和相关支撑技术。然后,对上述建模技术体系中的关键技术展开了研究,包括三维人脸建模、三维皱纹建模、三维个性化发型建模和个性化定制人像产品的顾客满意度评价技术。 三维人脸建模方面,提出一种基于通用模型和曲面约束变形的人脸建模方法,包括基于通用模型的个性化人脸宏观变形和基于曲面约束变形的人脸模型优化。其中,基于通用模型的个性化人脸宏观变形包括二维特征点的定义和提取、三维特征点匹配以及通用模型的宏观变形;基于曲面约束变形的人脸模型优化方法包括基于点约束变形的人脸模型优化、基于线约束的非均匀变形的人脸模型优化和基于模板和曲面约束变形的耳朵模型优化。 三维皱纹建模方面,提出一种基于形状控制函数和自适应细分的皱纹建模方法。其中,三维皱纹生成时,提出三个形状控制函数,即控制皱纹截面形状的横截面形状控制函数(Cross-section Shape Control Function,CSCF)、控制深度方向变化的深度衰减函数(Depth Attenuation Function,DAF)和控制宽度方向变化的宽度衰减函数(Width Attenuation Function,WAF);并提出一种皱纹影响区域的定义方法和网格自适应细分技术,不仅保证了皱纹建模的效果,而且能提高建模效率。 三维个性化发型建模方面,提出一种基于体积和头发簇的个性化发型建模方法。其中,头发体积建模集成了基于点约束变形的体积生成方法、基于点自由变形的头发边缘修正方法、基于等距偏移的头发体积快速生成方法以及基于模板的头发体积生成方法;个性化头发簇建模主要采用了基于线约束的非均匀变形的方法。 针对所建立的人像模型,提出一种个性化定制人像产品的顾客满意度评价方法。通过研究个性化定制人像产品的顾客满意度需求分析和评价模式,建立了包括技术性、品位性、经济性、效率性、服务性的顾客满意度评价指标体系和顾客满意度评价模型;该模型把顾客满意度评价分为对虚拟产品的初评和对实际产品的总评,详细介绍了各项指标值的量化方法。 最后,将上述方法应用到个性化定制人像产品开发中,给出了三个综合应用案例:全部数控加工的三维人像产品、手工制作与数控加工一体化的真人产品以及个性化真人浮雕产品。案例结果表明,本文方法具有较好的建模效果,能为个性化定制产品开发提供一种理论方法和技术支持。
[Abstract]:The manufacturing industry is the pillar industry of the national economy. An important trend of the development of the manufacturing industry is to continuously develop the personalized customized products to meet the market demand of rapid development. At present, a kind of personalized customized product related to human image is produced, and it is gradually forming a potential huge new industry. One of the key problems of personalized customized products related to human image is that people are like three How to get a realistic three-dimensional human image model is the most challenging question in the fields of computer-aided design, computer graphics, and advanced manufacturing technology One of the most direct methods of three-dimensional modeling is to use a three-dimensional scanning device to obtain the three-dimensional point cloud of human image. Although the method is high in precision, the method has the advantages of high equipment cost, inconvenient scanning of the modeling object, and difficult scanning of the reflection of the light. The most important means at present is to use two-dimensional image reconstruction to reconstruct human image three-dimensional model, which has been widely used in virtual scenes such as character animation, film and television advertising, game production and so on, and the established model is mainly a pattern that can meet the visual effect. however, that three-dimensional model is to be applied to the product development of the personalized customization, and there are several difficult problems: 1) the established model cannot be directly used for product processing and manufacture; 2) the established model has a higher weight at the feature point. The structure precision, but the error is large at the non-characteristic region and the characteristic line, the model similarity cannot meet the development of the product, and 3) the customer satisfaction rating of a human image product is absent In view of the above problems, this paper mainly studies how to use two-dimensional image to set up a three-dimensional model of human image with high similarity, and its application in the product development of personalized customization, so as to provide a new theory for industry. Methods and technical support. The main contents of the study In the following aspects, firstly, the individual customized product is discussed, and the overall framework of the human image three-dimensional modeling technology system facing the personalized customization product is established, including the modeling. The key technologies and related supporting technologies are also studied, including three-dimensional face modeling, three-dimensional wrinkle modeling, three-dimensional personalized hairstyle modeling and personalized customization. The invention provides a method for modeling a human face based on a general model and a surface constraint deformation based on a general model and a surface constraint deformation, which comprises the following steps of: The human face model optimization based on the surface constraint deformation includes the definition and extraction of the two-dimensional feature points, the matching of the three-dimensional feature points and the macroscopic deformation of the general model based on the general model, and the face model optimization method package based on the surface constraint deformation. Face model optimization based on point-constrained deformation, face model optimization based on line-constrained non-uniform deformation and die-based model optimization Ear model optimization for constrained deformation of plate and surface. In three-dimensional wrinkle modeling, a new method is presented. In this paper, three shape control functions, i.e., cross-section shape control function (CSCF), and depth attenuation function (Depth) for controlling the change in depth direction are proposed in the generation of three-dimensional wrinkles. The width attenuation function (WAF) of the variation of the orientation function (DAF) and the control width direction is proposed. The definition method and the mesh self-adaptive subdivision technique of the wrinkle influence area are presented, not only The effect of the wrinkle modeling is proved, and the modeling efficiency can be improved. The three-dimensional personalized hairstyle modeling The invention provides a method for modeling a personalized hair style based on volume and hair cluster, wherein, the hair volume modeling integrates a volume generation method based on point constraint deformation, a hair edge correction method based on point free deformation, Volume rapid generation method and template-based hair volume generation method In this paper, the method of non-uniform deformation based on line constraint is mainly used for cluster modeling. According to the model, a method for evaluating the customer satisfaction degree of a personalized customized human image product is presented. By studying the customer satisfaction demand analysis and evaluation model of the personalized custom-made person image product, the method includes the technical, the grade, the economy, the efficiency. Customer satisfaction evaluation index system and customer satisfaction evaluation model for sex and service; this model divides customer satisfaction evaluation into virtual products In the end, three comprehensive application cases: the three-dimensional human image of all numerical control machining are given in the paper. Product, hand-made and NC machining integrated real-life products and individual real-life relief products. The case results show that the method of this paper has the advantages of
【学位授予单位】：重庆大学
【学位级别】：博士
【学位授予年份】：2011
【分类号】：TP391.41
本文编号：2358043
[Abstract]:The manufacturing industry is the pillar industry of the national economy. An important trend of the development of the manufacturing industry is to continuously develop the personalized customized products to meet the market demand of rapid development. At present, a kind of personalized customized product related to human image is produced, and it is gradually forming a potential huge new industry. One of the key problems of personalized customized products related to human image is that people are like three How to get a realistic three-dimensional human image model is the most challenging question in the fields of computer-aided design, computer graphics, and advanced manufacturing technology One of the most direct methods of three-dimensional modeling is to use a three-dimensional scanning device to obtain the three-dimensional point cloud of human image. Although the method is high in precision, the method has the advantages of high equipment cost, inconvenient scanning of the modeling object, and difficult scanning of the reflection of the light. The most important means at present is to use two-dimensional image reconstruction to reconstruct human image three-dimensional model, which has been widely used in virtual scenes such as character animation, film and television advertising, game production and so on, and the established model is mainly a pattern that can meet the visual effect. however, that three-dimensional model is to be applied to the product development of the personalized customization, and there are several difficult problems: 1) the established model cannot be directly used for product processing and manufacture; 2) the established model has a higher weight at the feature point. The structure precision, but the error is large at the non-characteristic region and the characteristic line, the model similarity cannot meet the development of the product, and 3) the customer satisfaction rating of a human image product is absent In view of the above problems, this paper mainly studies how to use two-dimensional image to set up a three-dimensional model of human image with high similarity, and its application in the product development of personalized customization, so as to provide a new theory for industry. Methods and technical support. The main contents of the study In the following aspects, firstly, the individual customized product is discussed, and the overall framework of the human image three-dimensional modeling technology system facing the personalized customization product is established, including the modeling. The key technologies and related supporting technologies are also studied, including three-dimensional face modeling, three-dimensional wrinkle modeling, three-dimensional personalized hairstyle modeling and personalized customization. The invention provides a method for modeling a human face based on a general model and a surface constraint deformation based on a general model and a surface constraint deformation, which comprises the following steps of: The human face model optimization based on the surface constraint deformation includes the definition and extraction of the two-dimensional feature points, the matching of the three-dimensional feature points and the macroscopic deformation of the general model based on the general model, and the face model optimization method package based on the surface constraint deformation. Face model optimization based on point-constrained deformation, face model optimization based on line-constrained non-uniform deformation and die-based model optimization Ear model optimization for constrained deformation of plate and surface. In three-dimensional wrinkle modeling, a new method is presented. In this paper, three shape control functions, i.e., cross-section shape control function (CSCF), and depth attenuation function (Depth) for controlling the change in depth direction are proposed in the generation of three-dimensional wrinkles. The width attenuation function (WAF) of the variation of the orientation function (DAF) and the control width direction is proposed. The definition method and the mesh self-adaptive subdivision technique of the wrinkle influence area are presented, not only The effect of the wrinkle modeling is proved, and the modeling efficiency can be improved. The three-dimensional personalized hairstyle modeling The invention provides a method for modeling a personalized hair style based on volume and hair cluster, wherein, the hair volume modeling integrates a volume generation method based on point constraint deformation, a hair edge correction method based on point free deformation, Volume rapid generation method and template-based hair volume generation method In this paper, the method of non-uniform deformation based on line constraint is mainly used for cluster modeling. According to the model, a method for evaluating the customer satisfaction degree of a personalized customized human image product is presented. By studying the customer satisfaction demand analysis and evaluation model of the personalized custom-made person image product, the method includes the technical, the grade, the economy, the efficiency. Customer satisfaction evaluation index system and customer satisfaction evaluation model for sex and service; this model divides customer satisfaction evaluation into virtual products In the end, three comprehensive application cases: the three-dimensional human image of all numerical control machining are given in the paper. Product, hand-made and NC machining integrated real-life products and individual real-life relief products. The case results show that the method of this paper has the advantages of
【学位授予单位】：重庆大学
【学位级别】：博士
【学位授予年份】：2011
【分类号】：TP391.41
【引证文献】
相关硕士学位论文 前1条
1 王媛媛;面向数控加工的三维人像模型相似度评价技术研究及应用[D];重庆大学;2011年
，本文编号：2358043
本文链接：https://www.wllwen.com/wenyilunwen/guanggaoshejilunwen/2358043.html
