Kite Aerial Photography, as the name suggests, involves rigging a camera up to a kite system and then using it to take aerial photographs. After seeing my previous post on photogrammetry with a camera attached to a helium-balloon, Dr. Mike Smith at Kingston University contacted me about research he has been conducting in the realm of KAP and photogrammetry. Along with Drs. Chandler and Rose, he recently published a paper in Earth Surface Processes and Landforms entitled "High spatial resolution data acquisition for the geosciences: kite aerial photography".
The paper is relevant for the mapping industry because it provides an overview of the aerial acquisition process, the photogrammetric processing, and then an accuracy assessment of the results. I'll start with the results: their methodology enabled the production of stereo pairs, digital elevation models, and stereo imagery. Furthermore, the stereo imagery was triangulated with an accuracy of roughly 10mm in planform against surveyed ground control points.
The methodology involved using a 6 megapixel Nikon D70 camera and collecting aerial photography at altitudes of up to 200 meters over three test sites in the UK. GCP targets and XYZ samplings for topographic modeling were measured with Leica Geosystems TPS1200 and TCA 1105 Total Stations.
All the photogrammetric processing was performed in LPS. This involved setting up an LPS Blockfile (a project file), adding the images, and subsequently running through the aerial triangulation process in LPS Core in order to produce stereo pairs. With oriented images, the LPS Automatic Terrain Extraction module could be used to generate a digital elevation model. Next, the oriented images along with the digital terrain could be used to produce digital orthophotos. The paper describes the process in a high level of detail, as well as an excellent evaluation and discussion of the results.
Here is an image of an orthophoto superimposed with terrain points (red = automatically extracted, blue = measured via total station):
And here is a perspective view of an orthophoto draped over a corresponding digital elevation model, with contours:
In my opinion it is a great looking product considering it was generated with a 6 megapixel SLR camera flown from a kite!!
So why is this relevant for the mapping business?
The study illustrates a great low-cost approach to localized (as opposed to wide area) mapping, which means it may very well be a viable option for applications ranging from mapping cultural heritage sites to localized studies on soil erosion and other environmental and natural resource mapping projects. It is significant because it represents a significant cost saving over the traditional helicopter-based approach. If I had any talent for flying big kites I'd give it a whirl, but for now I'll leave it to he pro's...
Wednesday, March 11, 2009
Photogrammetry Meets Kite Aerial Photography (KAP)
Subscribe to:
Post Comments (Atom)
6 comments:
I love this stuff. Here is a link to terrain reconstruction by an archeologist using a blimp, camera and photosynth.
http://www.evolutionbeach.org/2008/11/topography-generation-through.html
Greg - very cool! Do you have any insight into the methodology? The results are impressive..
yes I have been working on a project using similar techniqes in Egypt.
It's not relevant because the costs of post-processing the imagery quickly outweighs any cost savings versus flying an airplane.
In the US and UK, you are barred by air traffic regulations from flying high enough to capture a large enough extent to reduce your image count and post-processing costs.
In an academic environment where you can get a bunch of grad students to do the orthorectification, the process works nicely.
For projects where you have a really small extent, it works beautifully. But this is nothing new. Archaelogists have been using kites, balloons and fixed platforms for decades.
And I'm not just blowing smoke. Check out myMS thesis.
Ebwolf: I agree with you on KAP having been in use for decades and I also agree that the technique isn’t applicable for wide-area mapping, but I disagree with your assertion of irrelevance. I think you may not be doing an apples-to-apples comparison. For accurate results, you’d need to perform post-processing of the imagery regardless of the sensor platform. Any high-accuracy (e.g. centimeter level) mapping project is going to use a photogrammetric approach to accomplish this.
In your thesis you’ve employed a wholly different workflow by georeferencing your imagery to much lower resolution imagery. This means you're not accounting for any relief variation as well as relying on the accuracy of your source imagery (which may contain many errors in itself). To improve your results you would really need to implement the methodology discussed above. In your conclusions you’ve dismissed the idea of orthorectifying the imagery yourself because of software costs and stating that “most digital elevation models are only accurate to ten meters or greater”. Now if you qualified that statement by saying “most free DEMs….” I’d agree, but again if you used a photogrammetric approach you could create your own DEM from stereo imagery (where the accuracy would be a function of your camera, the ground control, the flying height, and your triangulation results). As for software: cost is always relative. In the commercial world the ROI for photogrammetry software is quite high, otherwise you wouldn’t see so many successful commercial mapping firms. In academia, most software vendors have reduced rate programs.
See here for a real-world example employing a similar methodology to the one used in the paper above (albeit ballon-based), funded by the EU and government of Greece: http://fiducialmark.blogspot.com/2009/03/update-photogrammetry-at-acropolis.html. Another application that I suspect may be profitable would be open-pit mining. The geographic area is quite small but photogrammetry is required for volumetric assessment.
If you are interested in more photogrammetry from kite and balloon photography, look at some of the work my research group has been doing for the last years:
http://dx.doi.org/10.1016/j.geomorph.2008.05.047
More publications listed in http://www.geo.uni-frankfurt.de/ipg/ag/ma/marzolff.html
Regards, Irene
Post a Comment