[Back]

J. Otepka:
"Features of Georeferenced Point Clouds: Definition and Computation";
Talk: Europen Lidar Mapping Forum (ELMF), Amsterdam, The Netherlands; 2013-11-10 - 2013-11-13.

Today´s laser scanners, but also and dense matching algorithms, produce huge point clouds suitable for a variety of applications in different fields. Beside the coordinates, each point has various attributes attached which originate from the measurement process itself (e.g. amplitude, echo width, echo number, RGB values). The point cloud can be further enriched by calibrating the raw measurements, but also, by extracting geometrical and statistical features within the local neighborhood of each point. Such an enhanced point cloud may then serve as a basis for deriving application relevant models, like digital terrain models, vegetation layer, building models, etc. The contribution summarizes and classifies point features that were suggested in literature. Features extracted from the local neighborhood are embedded in a scale space. The scale parameter defined by the extent of the used neighborhood is crucial for the aimed application. For operational exploitation, reliable and efficient algorithms, tools, and work-flows are required. Accordingly, state-of-the-art processing strategies for extracting point features are analyzed, also regarding their processing efficiency. Additionally, the flexible point cloud processing software OPALS is used to demonstrate the extracting of various point features on data sets from different measurement strategies as well as the assembly of entire work flows.