Research Area:
Monitoring of Settlement and Open Space Development

Automated object recognition from high resolution LiDAR data and aerial imagery

Source: R. Hecht/IOER

High-precision sensors, such as laser scanners (both air-borne and terrestrial) and digital aerial cameras with Inertial-Measurement-Units (IMU) in combination with the possibilities of using Differential Global Positioning System (DGPS) reference stations produce very accurate and dense remote sensing data. Such data allow the development of spatial databases for a variety of large-scale applications. In the present project such data is tested as a basis for the extraction of railway infrastructure objects. The analysis focuses on automated processing under consideration of profitability.

The main task of the research project is the implementation of developed methods into a large scale database. Subsequently, these methods will be tested on datasets that have been collected using rail-born laser scanners. Such data resemble terrestrial scans. The point cloud density is equally high; however, the size of rail-born point clouds is much larger due to the extent of the corridors. Thus, an efficient data handling is only possible using a spatial database. Initially, different existing algorithms from digital image processing, image segmentation and object recognition are implemented in the spatial database. The final step is the documentation and evaluation of the quality and efficiency of both the database and rail-born laser scanner point clouds.