|
Introduction
We present a streaming framework for seamless building
reconstruction from huge aerial LiDAR point sets. By storing
data as stream files on hard disk and using main memory
as only a temporary storage for ongoing computation, we
achieve efficient out-of-core data management. This gives
us the ability to handle data sets with hundreds of millions
of points in a uniform manner. By adapting a building modeling
pipeline into our streaming framework, we create the
whole urban model of Atlanta from 17.7GB LiDAR data
with 683M points in under 25 hours using less than 1GB
memory. To integrate this complex modeling pipeline with
our streaming framework, we develop a state propagation
mechanism, and extend current reconstruction algorithms
to handle the large scale of data.
Pipeline
The input of our system is a set of aerial LiDAR point clouds capturing the surface geometry of certain urban area. Our system contains five modules as shown in the following figure:
- Pre-processing (Finalizer) : inserts finalization tags (markers that indicate spatial coherence) into the data, and produces a point stream.
- Classification (Classifier) : classifies noise and vegetation points from building and ground points.
- Segmentation (Splitter) : segments single building patches from the building and ground points. Two streams are generated - a point stream with geometry information of all ground points and a building stream which consists of individual building patches.
- Terrain generation (Terrain Generator) : converts the ground point stream into a terrain model by scan-convertion and interpolation.
- Building reconstruction (Modeler) : creates one polygonal building model for each building patch. A plane-based method is used in this step. A further extension introduces a more robust method called 2.5D dual contouring, visit the project page for more information.
Software
Unfortunately, the software is not published for free. If you are interested, contact Qian-Yi Zhou (qianyizh@usc.edu) and Ulrich Neumann (uneumann@graphics.usc.edu).
Publications
A Streaming Framework for Seamless Building Reconstruction from Large-Scale Aerial LiDAR Data
IEEE CVPR 2009, Qian-Yi Zhou and Ulrich Neumann
Links: Paper, Video, Poster
Fast and Extensible Building Modeling from Airborne LiDAR Data
ACM GIS 2008, Qian-Yi Zhou and Ulrich Neumann
Links: Paper, Slides
Related Links
|
