Andreas Wichmann

Matching Aerial Images to 3D Building Models Using Context-Based Geometric Hashing

A city is a dynamic entity, which environment is continuously changing over time. Accordingly, its virtual city models also need to be regularly updated to support accurate model-based decisions for various applications, including urban planning, emergency response and autonomous navigation. A concept of continuous city modeling is to progressively reconstruct city models by accommodating their changes recognized in spatio-temporal domain, while preserving unchanged structures. A first critical step for continuous city modeling is to coherently register remotely sensed data taken at different epochs with existing building models. This paper presents a new model-to-image registration method using a context-based geometric hashing (CGH) method to align a single image with existing 3D building models. This model-to-image registration process consists of three steps: (1) feature extraction; (2) similarity measure; and matching, and (3) estimating exterior orientation parameters (EOPs) of a single image. For feature extraction, we propose two types of matching cues: edged corner features representing the saliency of building corner points with associated edges, and contextual relations among the edged corner features within an individual roof. A set of matched corners are found with given proximity measure through geometric hashing, and optimal matches are then finally determined by maximizing the matching cost encoding contextual similarity between matching candidates. Final matched corners are used for adjusting EOPs of the single airborne image by the least square method based on collinearity equations. The result shows that acceptable accuracy of EOPs of a single image can be achievable using the proposed registration approach as an alternative to a labor-intensive manual registration process.

A Data Model for the Interactive Construction and Correction of 3D Building Geometry Based on Planar Half-Spaces

3D city models of large areas can only be efficiently (re-)constructed using automatic approaches. But since there is always a certain number of buildings where the automation fails, there is a need for interactive construction and correction tools. These tools should ideally use the reconstruction results as input, so that the amount of manual labor is minimized. However, automatic 3D building reconstruction approaches make use of different solid modeling techniques that are not all suitable for interactive modeling purposes. One such representation is half-space modeling that exhibits several advantages for the automatic (re-)construction of 3D building models (from segmented point clouds). Because planar half-spaces are infinite entities that are usually represented as mathematical inequality equations, it is difficult to design an interactive modeling system that allows their direct manipulation. In this paper, we propose an interactive modeling concept specifically for 3D building geometry based on a half-space kernel. Following from it, a special-purpose object-oriented data model is developed that hides the kernel under a layer of parameterized primitives and boundary representation (B-rep) that give semantic meaning to building elements and is thus better comprehensible to human users.

Smooth transformations between generalized 3D building models for visualization purposes

The generation of highly detailed 3D city models has significantly matured over the last couple of years. Especially for visualization purposes, these models are needed at several levels of detail (LOD) in fine granular steps that also allow for a smooth transformation from one level to the next to avoid pop-up artifacts. Several methods have been proposed for the automatic generalization of 3D building models that simplify the geometry of given boundary-based models at discrete LODs. In this paper, we present an approach to generate a transformation sequence between two discrete LODs with the purpose to later build a continuous LOD sequence. The method is based on mesh simplification techniques that are extended to a restricted use of edge collapse and vertex move operations applied on triangle meshes augmented by additional breaking lines. The validity of the approach is demonstrated on a case study that is close to reality.