视频数据空间化方法研究

项目名称： 视频数据空间化方法研究

项目编号： No.41201417

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 地理学

项目作者： 吴勇

作者单位： 福建师范大学

项目金额： 25万元

中文摘要： 视频作为一种常见的媒体，本身蕴含着丰富的空间和属性信息，不仅获取方便，而且表达的地理空间具有很强的真实感。现今GIS中视频常作为空间对象的一种属性进行存储和调用，使得视频没有得到充分利用。本项目拟通过对同步模型、耦合技术的研究，解决音视频信息与定位信息的融合问题，形成可定位视频；面向可定位视频，建立其与地理实体之间的匹配模型，实现视频的自动分割，在此基础上构建视频帧的视域模型，从视频帧、片段不同粒度层面建立高效的时空索引机制，实现基于时空的视频检索；构建可定位视频与三维场景模型间的映射关系，研究视频中实体目标帧间追踪机制,标注布局、调度策略，实现视频自适应标注算法；在上述理论研究与实验分析的基础上，构建视频GIS原型平台。研究不仅拓展数据采集方法和地理空间的表达方式，丰富地理信息科学的理论与方法，为公众提供一个全新的信息互动平台，也将在应急指挥、快速反应救援、实景旅游等领域发挥重大作用。

中文关键词： 可定位图像/视频；基于位置图像/视频检索；视频分割；视频传感器覆盖；相机标定

英文摘要： Videos can be seen as an objective presentation of all elements throughout our living environment, spatial and/or no-spatial attributes. However they are just treated as a property of the spatial object to store and recall, which results in the absence of the interpretation and analysis their contents. Therefore, this project is divided into three interrelated contents based on the main line of video "Integration, Management, Visualization". Firstly, we define the new synthetic media as the locatable video, which is the integration of audio, video and locating information by means of the synchronization model, the coupling technique. Secondly, we plan to realize the automatic segmentation algorithm based on the relationship between the locatable video and the geographic entity, such as the road network. On this basis, the locatable video retrieval is implemented by developing the vision model of video frame and the time-space based index of video. Thirdly, we develop a tracking mechanism for entity in video frames, method of scheduling and labeling the layout, and then realize the adaptive tagging algorithm. Above theoretical and experimental analysis, the prototype system is developed. The study would expand the method of capturing GIS data and the way of expressing geographic space, enrich the theory and metho

英文关键词： The Locatable Image/Video；Location-Based Image/Video Retrieval；Video Segmentation；Camera Coverage；camera calibration