Building acquisition, textured building models, integration of GIS-technology and image analysis

该项目的目的是开发模型、方法和工具，用于半自动、交互式地改进3D城市模型。从LOD2 （level of detail 2）的模型开始，由简单的建筑块和屋顶结构组成，细化到LOD3的模型，即带有地面边缘、楼梯、阳台、窗户、门和纹理的详细建筑模型。重点是建筑立面和数字立面模型之间的过渡，即所谓的建筑领。主要组件是一个三维空间数据库，它集成了矢量建筑模型和空中和地面光栅图像。精心设计的用户界面支持图像解释中歧义的交互式解决方案。对于几何和纹理的整合，将开发一个评估函数，以平衡模型和图像之间关于视觉合理性的不可分割的冲突。在第一阶段，我们开发了一个本体、工作流程场景、通用模型和程序，用于重建楼梯作为地形和建筑之间的主要联系。在第二阶段，我们将重点放在建筑领的更一般的部分，以及与交互系统的集成上，特别是使用随机语法、贝叶斯网络，可能还有马尔可夫随机场和蒙特卡罗技术来可靠地恢复基本结构。该方法的新颖之处在于：(1)高分辨率目标模型的规范；(2)从地理参考点云中重构实例化目标模型并根据模型规范进行泛化。(3)综合使用随机属性语法和图形模型进行图像解释。

The aim of the project is to develop models, methods, and tools for the semi-automatic, interactive refinement of 3D city models. Starting from models of level of detail 2 (LOD2), consisting of simple building blocks and roof structures, the refinement leads to models of LOD3, namely detailed building models with ground edges, stairs, balconies, windows, doors and textures. Focus is on the transition between building facades and the digital elevation model, the so-called building collar. The main component is a 3D spatial database which integrates vector building models and aerial and terrestrial raster images. A carefully designed user interface supports the interactive resolution of ambiguities in the interpretation of the images. For the integration of geometry and texture an evaluation function will be developed which balances the inextricable conflict between model and image with respect to visual plausibility. In the first phase we have developed an ontology, work flow scenarios, generic models and procedures for reconstructing stairs as the main link between the terrain and a building. In the second phase we focus on more general parts of the building collar, on the integration into an interactive system, especially using stochastic grammars, Bayesian networks, possibly also Markov-Random fields and Monte Carlo techniques for a reliable recovery of essential structures. The novelty of the approach is (1) the specification of high resolution object models, (2) the reconstruction of the instantiated object model from geo-referenced point clouds and its generalization according to the model specification. (3) the integrated use of stochastic attribute grammars and graphical models for image interpretation.