基于结构分析的建筑模型合成方法研究

发布时间：2018-12-12 14:12

【摘要】：数字几何模型已被公认为作为继数字音频、图像与视频后的新一代数字媒体,已在工业制造、数字娱乐等诸多领域得到广泛应用,可共享的数字几何模型数量更是在以指数方式快速增长,如何实现已有模型资源的有效重用就成为该领域面临的重要课题之一。基于样本的建模技术利用已有三维样本来合成新模型,不仅可以直观确定所建模型样式,还可以简化和加速建模过程,为模型资源的重用提供了一种有效途径,己成为数字几何处理领域的热点研究课题之一。本文以建筑模型为研究对象,采用基于样本建模方法,对其所涉及的模型结构分析以及后续的模型合成二大关键技术进行了系统的研究与探索。并在此基础上设计和实现了相应的原型系统。论文的主要工作成果体现在以下三个方面：1.提出了一种基于变换空间的三维建筑模型结构发现方法。该方法将求取输入模型上规则结构重复模式的问题转化成先求取平移变换空间中的网格规律模式再反推的问题,可以在三维建筑模型重复构造单元形状、大小等方面的先验信息未知的前提下,同步进行重复模式估计和平移重复构造单元检测,进而实现模型结构的发现。实验表明：该方法能够快速获得建筑模型中的平移重复构造单元,为后续模型合成奠定扎实的研究基础。2.提出了一种基于建筑单立面合成与重组的模型合成方法。该方法利用大多建筑模型是由各单独立面拼接组合的构造特点,将建筑模型合成问题转化为各建筑单立面分别进行合成后的组合拼接问题,进而可实现便捷的建筑模型合成。实验表明：该方法能利用单个样本或多个样本快速高效的生成风格迥异的多种建筑模型,为模型的优化合成问题提供了新的研究思路。3. Building-Synthesis系统设计与实现。该系统整合了本文所提出的三维建筑模型结构发现与基于建筑单立面合成与重组的模型合成方法,不仅可以给专业人士提供新的建模手段,而且还可以给非专业人士提供相应的建模体验。
[Abstract]:Digital geometry model has been recognized as a new generation of digital media after digital audio, image and video. It has been widely used in many fields such as industrial manufacture, digital entertainment and so on. The number of shared digital geometric models is growing exponentially. How to effectively reuse the existing model resources has become one of the important issues in this field. The modeling technology based on samples can make use of the existing 3D samples to synthesize the new model, which can not only determine the model style intuitively, but also simplify and accelerate the modeling process. It provides an effective way for the reuse of model resources. It has become one of the hot research topics in the field of digital geometry processing. In this paper, the architecture model is taken as the research object, and the two key technologies involved in the model structure analysis and subsequent model composition are systematically studied and explored by the sample-based modeling method. On this basis, the corresponding prototype system is designed and implemented. The main achievements of the thesis are as follows: 1. In this paper, a method of structure discovery of three-dimensional building model based on transformation space is proposed. In this method, the problem of obtaining the regular structure repetition pattern on the input model is transformed into the problem of obtaining the grid regular pattern in the translational transformation space and then the problem of backward inferring, which can be repeated in the three-dimensional building model to construct the shape of the unit. On the premise of unknown priori information such as size and size, repeat pattern estimation and translation repeat construction unit detection are carried out synchronously to realize the discovery of model structure. The experimental results show that this method can quickly obtain the translational and repetitive structural units in the building model, and lay a solid foundation for the subsequent model synthesis. 2. A model synthesis method based on building single facade synthesis and reconstruction is proposed. This method takes advantage of the structural characteristics that most building models are composed of individual facades, and transforms the problem of building model synthesis into the combinatorial problem of each building with single facade, which is convenient for building model synthesis. The experimental results show that this method can make use of a single sample or multiple samples to generate a variety of architectural models with different styles quickly and efficiently, and provides a new research idea for the optimization and synthesis of the model. 3. Design and implementation of Building-Synthesis system. The system integrates the 3D building model structure discovery method and the model composition method based on building single facade synthesis and recombination, which can not only provide a new modeling method for professionals. And can also provide the corresponding modeling experience for lay-professionals.
【学位授予单位】：南京大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TU31;TP391.41

【相似文献】