【作者】 吴江；

【导师】 步丰林；

【作者基本信息】 上海交通大学 ， 软件工程（专业学位）， 2015， 硕士

【摘要】 在很多专业领域中,三维模型是知识的载体并且能够产生巨大的经济价值。现在三维模型的建模工具已经相对成熟,不少领域都有各自的行业标准,近些年研究的热点开始转向如何复用和检索已有海量模型。如今用户对模型检索的要求变得越来越高,检索不再局限于关键字检索还要求系统能够应对复杂概念的检索,但是三维模型在高层次概念的语义缺失,使得现有的模型复用和检索的解决方案无法满足工业的需求。随着越来越多的领域涉及到三维场景建模过程中,传统的单台计算机设计模式受到严峻的挑战。如何在利用互联网技术使得多领域不同地域的人共同参与到同一个场景建模项目中也是迫切需要解决的问题。针对以上问题,本文提出基于本体的三维场景建模管理框架,支持多领域多人的三维场景建模,将多领域的领域知识充分地利用到场景建模过程中来,确保场景建模结果符合领域专家的初始设定,保证三维模型/场景的语义一致性,提高场景建模的效率以及三维模型的复用性。本文的具体工作内容如下:第一,提出三维模型场景建模管理框架。此框架通过利用本体的特性保证了三维模型语义的一致性,可以满足多领域专家参与到同一个场景建模工程中的需求,并且提供检测机制提供反馈,框架可以支持整个设计流程。第二,提出三层本体模型和语义约束模型。三层本体包括针对三维模型外观特征的通用本体,代表专业领域三维模型的领域本体,面向场景建模过程中语义约束的场景本体。在三层本体中,领域本体实例与通用本体实例的绑定代表一个概念模型,而场景本体实例代表一个语义约束模型。本文不但给出三层本体的形式化定义,还给出构建方法,概念模型的语义信息绑定方法,场景语义约束的表示方法,以及三层本体之间的关系。第三,设计针对三层本体的本体存储结构以及在网络环境中基于本体的约束检测机制。在模型语义方面,阐述三维模型与领域本体实例的映射方式。在语义约束方面,给出了模型语义信息和场景约束信息的获取过程,设计了场景约束检测流程。通过场景的语义约束检测功能,建模人员可以知道自己的模型场景是否符合规定以及违规的地方。此外,我们加入语义缓存模块来提高网络环境下大量模型的语义信息查询的响应时间,改善系统性能。最后,本文展示基于三层本体的场景建模系统原型,选取室内装潢为例,构建本体信息,验证模型的语义获取和语义标注功能,以及模型场景的约束检测功能。结果显示,本系统可以保证模型语义的一致性,支持场景约束验证,具有一定的实用价值。更多还原

【Abstract】 In many professional field, three-dimensional model is the carrier of knowledge and have an enormous economic value. In recent year, three-dimensional modeling tool become relatively mature and have many solutions and standards, how to reuse and retrieve existing models become a hot research. The demand of model retrieval become increasingly complex. The search is no longer limited to a keyword. In addition, we need the search system can deal with complex conceptal search requests. However, in the high-level aspect of concept, the semantic absence of three-dimensional model make the existing solutions of reuse and retrieve model can’t meet needs of industry. Meanwhile, a growing number of areas are related to the three-dimensional scene modeling process, traditional single computer design model is being challenged. It is become an urgent problem to be solved that how to use Internet technology to make people from different field and region to participate the same scene modeling project.To solve the above problem, a three-dimensional scene modeling management framework based on ontology has been proposed in this paper. This framework can support three-dimensional scene modeling with Multiplayer from multi-domain, ensure the semantic consistency of 3D model/scene, improve the accessibility and reusability of 3D model, make full use of domain knowledge from various fields in the process of scene modeling, make sure that scene model result meets the init expectations of domain experts. The main work of paper are shown as follows:(1) A three-dimensional model scene model management framework is proposed. This framework can ensure semantic consistency of 3D model by use the characteristics of ontology, meet the need that multi-domain experts participate in the same scene modeling project, provide the detection mechanism to get feedback, which ensures the integrity of the scene modeling process.(2) Three-layer ontology model and semantic constraint model are proposed. The Three-layer ontology model including the general ontology for 3D model appearance characteristics, the domain ontology for representing the 3D model in some domain, the scene ontology for semantic constraint in scene modeling process. In the three-layer ontology, the domain ontology instance which is binding with general ontology represents a conceptual model and the scene ontology instance represents a scene constraint model. This paper not only gives the formal definition, but also gives their construction method. Beside, we give the semantic information binding method of concept model, the representation of scene semantic constraint, as well as the relationship between the three-layer ontology.(3) We design the storage structure for three-layer ontology and the constraint detection mechanism based on ontology in the Internet environment. In the aspect of 3D model semantic information, we give the mapping mode between 3D model and domain ontology. In the aspect of semantic constraint, we give the model semantic information and scene constraint information acquisition process and design the scene detection process. By scene semantic constraint detection, modeler can knows whether the scene model has any violations and the violations if there are exist. Meanwhile, we add semantic cache module to the system to shorten the semantic query response time of a large number model in the Internet environment, which improves the system performance to a certain extent.A scene modeling prototype system is implemented to verify the effectiveness of our method. We choose upholstery as study case, a related ontology is constructed as model and scene information. Then, we use the system to verify the 3D model semantic annotation and acquisition function, and the scene model constraint detection function. The result shows the system is work well as our expert and our method has a practical application value.更多还原