Las Vegas Reconstruction
The Las Vegas Reconstruction Toolkit provides an open source C++-library of several algorithms for polygonal map generation. It is licensed under the GNU Public Licence. Currently Linux and Mac OS are supported, a Windows port is in development. The library structure is strictly modular, so that new features can be integrated easily. It comes with a detailed API documentation and example programs for surface reconstruction and a viewer. The reconstruction components can roughly be divided into three main components: Surface reconstruction, mesh optimization and texture generation. The main focus for the reconstruction is to create surface representations that can be used in robotic contexts like localization (via ray tracing) or scene interpretation (using spatial reasoning), rather than other constraints that are discussed in the computer graphics community like triangle quality, well formed topology in meshes, i.e. no t-edges, and so on. Our software primarily exploits the inherent planar scene structure that is found in many robotic applications.
- High-performance Marching Cubes based surface reconstruction from unorganized points.
- Mesh optimization based on planar clustering.
- Texture generation from colored point clouds or laser scanner intensity values.
- Robust point normal estimation for sparse point clouds.
- Parallelized implementation for multi core systems.
- Hole filling in the reconstructed meshes.
- Point cloud coloring from reference clouds.
- Support of various file formats.
- Video export of flights through point clouds and meshes.
Provided sample tools:
- Surface reconstruction
- Viewer application
- Point cloud reduction tool
- Point cloud colorization tool
- File format conversion tool
- Outlier removal and MLS projection (from PCL)