地理立体视频数据模型与验证

发布时间：2018-05-05 02:35

本文选题：立体视频 + 视频GIS　；参考：《北京建筑大学》2016年硕士论文

【摘要】：在"大数据"时代背景下,地理信息的应用发展正在走向"信息高速路"。随着GIS应用领域的不断扩充和细分,GIS的应用模式也在不断的改变。近些年,"视频+GIS"的应用模式成为一种新的应用趋势,不论是在城市管理方向、安防监控方向、交通监测方向及其他道路方面业务等领域,逐渐青睐于对这种应用模式的探索研究。通过视频采集影像与空间数据进行融合,在原有"数字化"的基础上,让数据"可视化",使数据内容更加立体、应用体验更加丰富。本文将在视频GIS的基础上,引入当下流行的"立体视频"概念,并在"立体视频+GIS"背景下,就数据采集处理与集成、数据组织及实施方案、数据描述与检索方法和数据信息冗余优化方法等方面进行了研究与实现,具体工作内容如下:1.地理立体视频数据的采集处理与集成首先,阐述立体视觉理论,包括:立体成像原理、视差理论及视差估计、数据采集处理过程中涉及的三种坐标系(世界坐标系、摄像机坐标系、图像坐标系)及三种坐标系间的关系,为后续数据采集及处理工作作好理论铺垫;其次,设计移动数据采集平台,在同时采集影像视频数据、空间数据基础上实现数据采集、处理、存储同步进行;最后,就数据处理方法和流程进行介绍,完成数据模型建立的准备工作。2.地理立体视频数据概念模型、逻辑模型的建立通过对地理立体视频数据的分析,建立数据模型约束条件(时空耦合性、扩充性、符合MPEG-7标准),基于约束条件,分别对包含的多种数据类型(视频数据、空间数据、线性参考、属性数据、元数据、语义等)分别建立子数据模型,最后通过描述各自子模型间的关联性,最终确定完整的地理立体视频概念数据模型。在概念数据模型基础上,从逻辑角度出发,精简出在本实验中较为重要的14个元素,对其分别设计数据存储表,讨论数据组织与数据结构问题,为实现视频对象的数据检索提供依据。3.地理立体视频数据描述与检索首先提出需要解决数据模型进行语义描述的问题,并引入"MPEG-7多媒体内容描述标准接口"的概念,对MPEG-7描述框架、内容描述工具作简单介绍,并基于Object DS、EventDS、SemanticPlaceDS和SemanticTimeDS等工具对数据语义进行描述,并在后续根据数据类型的不同,分别采用基于语义关系、线性关系和空间拓扑关系三种不同的检索方法实现视频数据的检索。4.地理立体视频数据信息冗余及编码方法从地理立体视频数据在处理数据实时传输处理中的数据冗余对网络传输和通信信道造成的压力入手,提出了对地理立体视频数据进行数据编码的假设,结合目前主流的H.264编码框架,实现基于对象的地理立体视频编码,为实现数据实时传输处理提供服务。
[Abstract]:Under the background of big data, the application and development of geographical information is heading for the information highway. Along with the continuous expansion of GIS application field and the application mode of subdivision GIS is changing constantly. In recent years, the application mode of "Video GIS" has become a new application trend, whether in the direction of urban management, the direction of security monitoring, the direction of traffic monitoring and other areas of road business, etc. Gradually favor the application of this mode of exploration and research. Through the fusion of video capture image and spatial data, on the basis of the original "digitization", the data can be "visualized", so that the data content will be more three-dimensional and the application experience will be more abundant. On the basis of video GIS, this paper introduces the concept of "stereo video", and under the background of "stereoscopic video GIS", the data acquisition, processing, data organization and implementation scheme are discussed. The methods of data description and retrieval and redundancy optimization of data information are studied and implemented. The detailed work is as follows: 1. Firstly, the theory of stereoscopic vision, including stereo imaging principle, parallax theory and parallax estimation, three kinds of coordinate systems (world coordinate system) involved in data acquisition and processing are expounded. The relations between camera coordinate system, image coordinate system) and three coordinate systems lay a good foundation for the subsequent data acquisition and processing. Secondly, the mobile data acquisition platform is designed to collect video data at the same time. On the basis of spatial data, data acquisition, processing and storage are carried out synchronously. Finally, the method and flow of data processing are introduced to complete the preparation of data model. 2. The concept Model of Geography Stereo Video data, the logical model is built through the analysis of the geographic stereo video data, the constraint conditions of the data model (spatiotemporal coupling, extensibility, accord with the MPEG-7 standard, based on the constraint condition) are established. Subdata models are established for various data types (video data, spatial data, linear reference, attribute data, metadata, semantics, etc.). Finally, the complete geographic stereo video conceptual data model is determined. On the basis of conceptual data model, 14 important elements in this experiment are simplified from the point of view of logic, and the data storage tables are designed respectively, and the problems of data organization and data structure are discussed. Provide basis for data retrieval of video object. 3. 3. Firstly, the problem of semantic description of data model is proposed, and the concept of "MPEG-7 multimedia content description standard interface" is introduced. The MPEG-7 description framework and content description tools are briefly introduced. The data semantics is described based on the tools of Object DSV event DS and SemanticTimeDS. According to the different data types, three different retrieval methods based on semantic relation, linear relation and spatial topological relation are used to realize the retrieval of video data. The redundancy and coding method of Geography Stereo Video data start with the pressure on network transmission and communication channel caused by data redundancy of geographic stereo video data in processing data real-time transmission processing. The assumption of data coding for geographic stereo video data is put forward. Combining with the current mainstream H.264 coding framework, the object based geographic stereo video coding is realized, which provides a service for real-time data transmission and processing.
【学位授予单位】：北京建筑大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：P208;P209

【参考文献】