Feature preserving decimation of urban meshes

Abstract

Commercial buildings as well as residential houses represent core structures of any modern day urban or semi-urban areas. Consequently, 3D models of urban buildings are of paramount importance to a majority of digital urban applications such as city planning, 3D mapping and navigation, video games and movies, among others. However, current studies suggest that existing 3D modeling approaches often involve high computational cost and large storage volumes for processing the geometric details of the buildings. Therefore, it is essential to generate concise digital representations of urban buildings from the 3D measurements or images, so that the acquired information can be efficiently utilized for various urban applications. Such concise representations, often referred to as “lightweight” models, strive to capture the details of the physical objects with less computational storage. Furthermore, lightweight models consume less bandwidth for online applications and facilitate accelerated visualizations. In this thesis, we provide an assessment study on state-of-the-art data structures for storing lightweight urban buildings. Then we propose a method to generate lightweight yet highly detailed 3D building models from LiDAR scans. The lightweight modeling pipeline comprises the following stages: mesh reconstruction, feature points detection and mesh decimation through gradient structure tensors. The gradient of each vertex of the reconstructed mesh is obtained by estimating the vertex confidence through eigen analysis and further encoded into a 3 X 3 structure tensor. We analyze the eigenvalues of structure tensor representing gradient variations and use it to classify vertices into various feature classes, e.g., edges, and corners. While decimating the mesh, fea ture points are preserved through a mean cost-based edge collapse operation. The experiments on different building facade models show that our method is effective in generating simplified models with a trade-off between simplification and accuracy.