|Digital surface model|
Although commercial software for the automatic generation of digital surface models (DSM) by image matching is available for more than 20 years, the matching results have never been such promising as today. Compared to scanned analogue aerial images, digital photogrammetric frame cameras like our UltraCamX or UltraCamEAGLE feature a high image dynamic as well as an excellent signal-to-noise ratio. These improvements are responsible for a higher accuracy, better reliability and density of automatic image matching. Several investigations like the DGPF-project on digital photogrammetric camera evaluation demonstrated the quality and the potential of generating elevation models from digital imagery. Flight configurations with 8 cm ground sampling distance (GSD), 80% forward overlap and 60% side lap are comparable to LiDAR data in terms of the achievable vertical accuracy (see PFG 2010, No. 2).
The use of digital aerial imagery for automatic DSM and derivate terrain models (DTM) contains the following advantages:
The high quality from digital airborne cameras allows the generation of 3D point clouds for applications that has been exclusively managed by LiDAR measurements in the past. Nowadays elevation data from digital images could be used for many tasks like 3D building models, roof shapes, canopy models, true-orthophotos or 3D-landscape visualizations. In the above figure you can see the results of a digital image matching with Match-T, showing a small part of an area in Kurdistan. The images were taken by our UltraCamX with a GSD of 10cm. The block configuration is 60% forward overlap and 30% side lap. The example shows the orthophoto, the corresponding DSM as well as a colorized DSM within a four sq.km test area. The DSM has been derived from the original 3D-point cloud and was interpolated to a 50 cm regular grid. Obviously visible is the high level of detail, represented by approximately sixteen million (!) height-points. Breaklines that could further improve the quality of the DSM are not used within this example.
Of course at the current (2011) state of the art there are still some restrictions in generating DTM’s by image matching. Difficult regions like urban areas are requiring complex filter algorithm. Also leafiness is a special problem. But don't forget, we are at the beginning of promising developments. Software and algorithms will be further improved. The potential of digital imagery is not fully used at this time. Future comes!