Auto 3-d complex building detection from satellite imagery

从高分辨率光电图像中提取三维建筑物几何信息正成为众多地理空间应用中的一个关键因素。事实上，生成3D城市模型是各种应用的要求，例如城市设计的空间分析、军事模拟和特殊地理位置的现场监控。多年来开发的用于3D建筑重建的几乎所有操作方法都是半自动的，在这种方法中，熟练的人工操作员参与建筑实例的3D几何建模，这是一个耗时的过程。此外，这种方法通常需要立体图像对、图像序列或特定地理位置的激光扫描来从图像中提取3D模型。最后，在目前的技术下，3D几何建模阶段的特征可能是提取精度较低的3D建筑模型。本研究的目标是利用单目全色航空/星载图像来处理任意视点变化、杂波和遮挡的成像，从而实现全自动的3D结构检测。该系统将被设计为处理复杂的形状和干扰引起的复杂问题，这是由于城市建筑非常接近而引起的。从遥感图像中提取几何信息的一致、可靠的系统有许多应用。其中一些包括：城市规划和机场危险分析、民事紧急情况缓解、自然灾害管理(洪水、地震和山体滑坡)、军事监视(积极参与武力、反恐和维持和平措施以及统计地理定位(如健康、犯罪和过去的自然灾害)。

The extraction of 3D geometric building information from high-resolution electro-optical imagery is becominga key element in numerous geospatial applications. Indeed, producing 3D urban models is a requirement for avariety of applications, such as spatial analysis of urban design, military simulation, and site monitoring of aparticular geographic location. Almost all operational approaches developed over the years for 3D buildingreconstruction are semi-automated ones, where a skilled human operator is involved in the 3D geometrymodeling of building instances, a time-consuming process. Further, such approaches usually require stereoimage pairs, image sequences, or laser scanning of a specific geographic location to extract the 3D models fromthe imagery. Finally, with current techniques, the 3D geometric modeling phase may be characterized by theextraction of 3D building models with low accuracy levels.This research aims for a fully automated 3D Structure Detection that copes with imaging of arbitrary viewpointvariations, clutter, and occlusion using monocular panchromatic air/spaceborne images. The system will bedesigned to deal with complications arising from complex shapes and interference due to the close proximity ofurban city buildings. There are numerous applications for a consistent reliable system of extracting geometricinformation from remotely sensed imagery. Some of these include: urban planning and airport hazard analysis,civilian emergency mitigation, natural disaster management (flooding, earthquakes, and landslides), militarysurveillance (active engagement of force, counter terrorism and peace keeping measures, and for statisticalgeographic localization (such as health, crime, and past natural disasters).