网格变形中的特征线提取与曲面重构研究
发布时间:2018-05-30 12:09
本文选题:网格变形 + 模型分块 ; 参考:《杭州电子科技大学》2011年硕士论文
【摘要】:’网格变形是三维模型处理领域中的重要组成部分,它是近几年来计算机图形学中的研究热点之一,很多研究成果已经得到了广泛的应用,如动画制作,影视广告,以及计算可视化等领域。网格变形大体可分为两类:一类是基于能量优化,另一类是基于插值重构。前者属于间接变形,速度较快、效果较好,但是变形不易控制,且适用范围较窄;后者则具有较强的操作性,算法简单实用,但效率不高而且容易失真。根据上述两类算法的优劣性,衍生出结合多种方法的混合变形技术,已成为近年来网格变形的发展趋势。本文针对插值重构方法中最流行的提取模型轮廓特征线来重构变形曲面的方法进行了相关研究,主要作了以下三个方面的工作。 第一,研究了形状直径函数(Shape Diameter Function,简称SDF)值的模型分块技术,并在此基础上实现了一种利用K-D树空间划分的网格模型分块方法。该方法利用K-D树的空间划分,通过它的每一层检测不同的属性(关键字)值来决定选择分枝的方向,使划分后的模型块与物体的实际结构相吻合。与传统的利用SDF值模型分块方法相比,本文给出的算法在一定程度上增加了模型分块的有效性,同时又为后期的骨骼线提取打下了良好的基础。 第二,通过基于K-D树空间划分的SDF值模型分块,给出了一种网格模型骨骼线和模型轮廓特征线的提取方法。模型轮廓特征线变形由于其自身提取困难,比较适合于给定特征线参数值的情况,且变形范围较窄。而单纯的使用骨骼线驱动变形又容易出现不自然的效果。本文通过把网格模型轮廓特征线结合到模型骨骼线上,让骨骼线变形的易于操纵、简单实用等优势和特征线变形保持细节的优势结合起来,增强网格模型的变形效果。 第三,提出了一种用轮廓线作为边界条件,然后利用偏微分方程来实现动态曲面重构方法。由于偏微分方程构造曲面简单易行,只需给定曲面的边界及其上的跨界导矢,即可生成光滑曲面。因此,本文通过在偏微分方程表示的曲面轮廓线中引入变形参数来代替点关联的方法,实现动态曲面重构。实验结果表明,本文方法所重构出的变形曲面自然连贯且客观真实,可应用于艺术曲面造型设计中。
[Abstract]:Grid deformation is an important part of 3D model processing field. It is one of the research hotspots in computer graphics in recent years. Many research results have been widely used, such as animation production, film and television advertising, And computational visualization and other fields. Mesh deformation can be divided into two categories: one is based on energy optimization, the other is based on interpolation reconstruction. The former belongs to indirect deformation, the speed is faster, the effect is better, but the deformation is not easy to control, and the scope of application is narrow; the latter has strong maneuverability, simple and practical algorithm, but low efficiency and easy distortion. According to the advantages and disadvantages of the above two algorithms, the hybrid deformation technology combined with many methods has become the development trend of grid deformation in recent years. In this paper, the most popular method of extracting the contour feature line of the model to reconstruct the deformable surface is studied in this paper. The main work is as follows. Firstly, the model partitioning technique of shape Diameter function (SDFs) of shape Diameter function is studied, and a mesh model block method based on K-D tree space is implemented. This method uses the space partition of K-D tree to determine the direction of branch selection by detecting different attribute (keyword) values in each layer of K-D tree, which makes the partitioned model block coincide with the actual structure of the object. Compared with the traditional block method using SDF value model, the algorithm in this paper increases the validity of the model to some extent, and lays a good foundation for the later skeleton line extraction at the same time. Secondly, by dividing the SDF value model into blocks based on K-D tree space, a method of extracting the skeleton line and the contour feature line of the mesh model is presented. Because of the difficulty of extracting the contour feature line of the model, it is more suitable for the given parameter value of the feature line, and the deformation range is narrow. But the simple use of bone lines to drive deformation is prone to unnatural effects. In this paper, the mesh model contour feature line is combined with the model skeleton line, so that the deformation of the bone line can be easily manipulated, simple and practical, and the advantages of keeping the detail of the feature line deformation can be combined to enhance the deformation effect of the mesh model. Thirdly, a dynamic surface reconstruction method using contour as boundary condition and partial differential equation is proposed. Due to the simple construction of surface by partial differential equation, the smooth surface can be generated only by giving the boundary of the surface and its cross boundary derivative vector. Therefore, in this paper, the dynamic surface reconstruction is realized by introducing the deformation parameter into the contour of the curved surface represented by partial differential equation instead of the point correlation method. The experimental results show that the deformed surface reconstructed by this method is naturally coherent and objective, and can be applied to the design of artistic surface modeling.
【学位授予单位】:杭州电子科技大学
【学位级别】:硕士
【学位授予年份】:2011
【分类号】:TP391.41
【参考文献】
相关期刊论文 前2条
1 王年;范益政;鲍文霞;韦穗;梁栋;;基于图割的图像匹配算法[J];电子学报;2006年02期
2 刘浩;廖文和;;基于Catmull-Clark细分的曲面重构[J];中国科学院研究生院学报;2007年03期
,本文编号:1955250
本文链接:https://www.wllwen.com/wenyilunwen/guanggaoshejilunwen/1955250.html