Rico Richter
Research Scientist
Hasso Plattner Institute
Potsdam University
Germany

Biography
Rico Richter is a research scientist with the computer graphics group of the Hasso Plattner Institute at the University of Potsdam, Germany. His research interests include processing and analysis techniques, scalable GPU-centric computing solutions as well as visualization and real-time rendering techniques for massive 3D point clouds.

Abstract
3D Point Cloud Analytics for Updating 3D City Models

Remote sensing technology such as LiDAR and image matching algorithms can be used for a periodical capturing of landscapes and urban regions. The resulting 3D point clouds commonly serve as the input for workflows and reconstruction algorithms to derive high-quality, geometrically complex and large-scale 3D city and landscape models. Due to a regular data acquisition (e.g., once a year) the efficient maintainance of these 3D models becomes an important task.

In this talk, we present concepts and techniques to update 3D city and landscape models based on 3D point cloud analytics. Change detection and classification techniques for time-variant 3D point clouds are used to detect, categorize, and quantify urban changes for buildings, vegetation, and terrain. GPU-based processing schemes and out-of-core algorithms allow us to cope with massive 3D point clouds featuring billions of points in short processing times.

As a case study, we present the results for an urban area of a city with an extend of 900 km² and approximately 80 billion points. The results show that large-scale, dense, and time-variant 3D point clouds can be used to establish continuation and maintaining workflows for 3D city and landscape models. We present our system infrastructure and point out new ways to manage, process, and analyze 3D point clouds in an automatic way as required by a growing number of applications and systems.