基于低空全景视频的地理环境模型构建与可视化技术研究

With the fast development of information industry in the past few years, geographic information are used in more and more fields, especially when army, police, fire fighters, traffic department and so on are put their eyes on some big expected events and nature disasters. It is desirable for surveying and mapping department to supply geographic information support in both 2D and 3D forms, which must be real-time, dynamic, constant, large range, measureable. These requests have excess the category of traditional fundamentally spatial 4D production..Therefore, this task takes the panorama videos screened by unmanned aerial vehicles as data source, and do some research about video technology, panorama technology and geographic information technology. We take out every frame image form video serial images, which is modeling into a panorama image with math elements. In the time axis, we establish relationship between video snippet and geographic spatial coordinate frame. Based on these work, we can get a panorama video geographic environment, which can be measureable and interfaced with, combined by geographic spatial data and panorama video. In the technology lay, we use the newest aero photography technology, computer graphics and images technology and geographic information technology, and realize integration of many subjects. This paper is very helpful for the development of visual geographic environment theory and it exploits a new research point.

近年来，随着社会信息化进程的加速，各领域对地理信息的需求越来越高，特别是军队、公安、消防、交通等部门应对重大突发事件、自然灾害的处理时，要求测绘部门能够提供实时、动态、连续、大范围、可量测、二维正射与三维透视相统一的地理信息的支撑，这些要求已大大超越了传统基础地理空间4D产品所涵盖的范畴。.因此，本课题将以无人飞行器等低空平台获取的全景视频作为数据源，以视频技术、全景技术、地理信息技术结合为切入点，将视频序列影像中的每一帧影像经过自动化建模处理形成具有空间数学基础的全景图像；在视频时间轴上，建立视频片断与地理空间坐标框架间的坐标映射关系，在此基础上构建一个地理空间数据与全景视频融合、具有可量测、交互控制等功能的全景视频地理环境。在技术层面上，应用了最新的航空摄影技术、计算机图形图像技术、地理信息技术，实现了多学科集成；在科学层面上有助于虚拟地理环境理论、技术的发展，开拓新的研究方向。

低空全景视频数据作为地理空间数据的一种新型数据源，具有获取方便、来源丰富、信息量大、现势性强等特点。将低空全景视频数据经过预处理，与精准的地理空间数据有机融合，构建基于全景序列影像的实时视频GIS，必将成为今后一段时期非常具有研究价值和应用前景的发展方向。本项目以基于全景序列影像的实时视频GIS为主要目标，突破了基于鱼眼镜头的序列影像纠正、地理全景视频的校准与标定、地理全景视频分割与全景序列影像数据与地理空间数据融合等关键技术，初步构建了基于低空全景视频的地理信息融合显示原型系统，本课题将为改变视频GIS的数据来源、构建方法与应用模式起到积极的推动作用，具有较强的理论意义和实用价值。