Point cloud data are important sources for 3D geo-information. The point cloud data sets are growing in popularity and in size. Modern Big Data acquisition and processing technologies, such as laser scanning from airborne, mobile, or static platforms, multi-beam echo-sounding have the potential to generate point clouds with millions or billions (or even trillions) of 3D points. Often these 3D data have one or more attributes associated with each point. This is especially true with the available and expected repeated scans of same area (the temporal dimension). These point clouds are too massive to be handled efficiently by common geo-ICT infrastructures. At the database level, initial implementations are available in both commercial and open source database products, illustrating the user need for point cloud support; e.g. Oracle spatial’s SDO_PC data type and PostgreSQL/PostGIS PCPATCH data type. Given the huge volumes of point cloud data, performance is critical, both with respect to data storage/ transfer size and to speed (for loading and querying). Therefore, it will be a real challenge to make an efficient implementation.

In this workshop, we present several methods to efficiently store and manage these massive point cloud data sets in a database. The way point cloud data is stored on the disk plays a major role in the query performance against these data sets. We will discuss several techniques including space filling curves based ordering to show the need for efficient organziation of data on the disk for good load and query performance. We discuss several storage models query models to show how databases can help efficiently manage these large volumes of data. The queries on point clouds often use non spatial attributes (time stamp, intensity, color, etc) to filter the data and hence databases are ideally suited for supporting such queries.

We also present some file based alternatives to compare and contrast the database solution for the point cloud data management.