Ground Control Points for aerial photographs - question

[Daniela M.]

Hello,

with our small UAV <5kg (incl. camera) we are planing to document an archaeological excavation site. The area is about 35 x 50m large. We use the photogrammetry software aspect3d (www.aspect3d.de) for computing the pictures. Right now we are in the testing phase and we realize that we need Ground Control Points for geo-referencing to achieve more accurate and faster results. Does anyone has experience working with GCP? How many GCP will we need for an area that size and how large should these be when flying about 15-20m high? Or the other way around: are there standard sizes for which we are to fly at a certain height? Where are these GCP available to buy (weatherproof) in the EU? Maybe there are free files existing on the net to print out our ourselves?

 

We (two fellow students and I) have been photographing just with tripod so fare but no GCPs. The results were pretty astounding (with objects small and medium). But with an entire area how would we go about taking additional pictures from the ground additionally to the photos from the UAV? How many GCPs and which size would we then need from the ground? (Sorry for the long and probably naive questions but we are just getting started and we will have only a small window to do the flight and ground photos - so we are a bit desperate).

 

Many thanks in advance!

[GeorgeBevan]

Daniela, 

 

Generally you will want to use a minimum of 4 GCPs in an aerial project like this. While 3 GCPs can effectively scale and orient your model there is no redundancy in case one of your measurements is incorrect, as sometimes happens, or is obscured in the photos. Ideally you would shoot in about 7 GCPs so that the software can actually improve on the accuracy of the survey (the inherent accuracy of photogrammetry, particularly in plan, will always be superior to that of a total station or even good GNSS receivers). These GCPs should be evenly, and randomly distributed throughout the area your are photographing. Avoid putting the GCPs in straight lines ("co-linear"). Are you using a total station or GNSS? If the former, will you be using a reflector pole or shooting the targets reflectorlessly?

 

Without knowing more about the camera and the software it is difficult to say how large the targets should be. You'll need to calculate the Ground Sample Size for your project, a value that is dependent on the camera, sensor size, focal length of the lens and flying height. Many software packages will detect targets automatically with right sort of targets. I couldn't determine what sort of targets Aspect3D uses from the online documentation. Perhaps the fastest solution would be purchase targets made for a total station that can be easily picked out from the air. There are also some LiDAR targets printed on rigid substrates that would work. Others have used targets used for archery and secured them to the ground or taped them to a heavier material so they don't blow away. I'd look for something at least 15cm in size given your flying height, but I can't be 100% sure. 

 

Hope that helps. 

[Taylor]

I developed an informal way of marking the locations of ground control points (GCPs) in Agisoft Photoscan that seems to work in cases where the targets are blurry or lack a well-defined center.  It seems to have improved the accuracy of the markers, so I thought I'd describe it here in case others find it useful.  This method has the advantages of being able to work with image data sets that are blurry or don't have a clearly marked center pattern (e.g., an absence of intersection of orthogonal lines, coded targets, or circle with a center), or lack sufficient resolution to clearly identify a pixel representing the center.   It seems to work well even in oblique images where the target has low contrast or is viewed from a low angle.  It could also be applied to markers for other features in images used for 3D digital reconstructions using photogrammetry.

 

Basically, the method involves looking for the centroid or weighted average of the pixels representing the target GCP.  I look for patterns of symmetry in the pixels representing the target, while paying attention to both color and intensity.  I liken this to the highlight detection algorithm in the RTIBuilder software, which finds the center of highlights on the reflective spheres.  I suppose that a similar algorithm could be written to detect arbitrary GCPs in Photoscan, which wouldn't require coded targets.  It requires only that the targets have a symmetric shape and  sufficient contrast to be identified visually against the surrounding terrain.  In many cases, the presence of a red and green fringe (chromatic aberration) around the edges of high-contrast targets helped to identify symmetries that could help locate the center of the targets.

 

As an example, this method allowed a series of GCP targets to be marked so the average projection error was limited to less than 0.25 pixels.  When the average projection error for a target was much higher than this value, the marker positions could easily be reviewed and the markers on images that contributed to the higher errors were removed or adjusted.  I found that it significantly improved the calculated error values.  Although the method as described here is subjective and could be said to introduce bias, I found that it helped me identify errors in my placement of the markers and correct them.  In this case, the targets "constructed of white sheets were deployed... and their positions recorded by dGPS to an estimated precision of 0.1 m."  (James, M. R. and Robson, S. [2012] Straightforward reconstruction of 3D surfaces and topography with a camera: Accuracy and geoscience application, J. Geophysical Res., 117, F03017, doi:10.1029/2011JF002289, accessed 14 November 2015). 

[GeorgeBevan]

Quick question, Taylor. When you say GCPs are you referring to markers used for scaling or markers with known 3D co-ordinates?

 

Centroiding is certainly a very handy way to digitize points. The accuracy, in theory, can be 0.01 pixels or lower. 

[Taylor]

Hi George,

 

I started using the technique when practicing on a set of publicly available aerial imagery that had GCPs surveyed using dGPS to 0.1 m, according to the source.  The GCPs were just described as "white sheets" with no target patterns or dimensions given, but they appeared as white squares or parallelograms when viewed from an angle in the images.  I noticed the red-green fringes from chomatic aberration around the white GCPs were helpful in defining symmetry I could use to find the centroid.  It seems to work even if the targets were a bit out of focus or seen in very low-angle views. 

 

I've also tried it also on markers for scaling, but the technique depends on the target pattern on the scale being marked; high-contrast squares or circles on the scale seem to work, but lines are tricky.  Once I've set the marker on two images, Photoscan will project the marker on other images, but I don't adjust the projected markers unless there's a good pattern visible.

[GeorgeBevan]

Very interesting! Could you share the link to the aerial data? I'd be interested to see what sort of error numbers can be obtained with your technique. 

[Taylor]

Here's the link (scroll down to the Reference section under "Volcanic edifice" to get the images and georeference data):  http://www.lancaster.ac.uk/staff/jamesm/software/sfm_georef.htm

 

I used just 5 of the GCPs to georeference the model and the errors averaged 0.186 to 0.261 pixels for 21-49 projections per GCP target.

 

I also tried using the "SfM-georef" software offered at the link, and it seems to install fine, but I couldn't get it to open my Photoscan (.psz) file, following the instructions.

[GeorgeBevan]

Thanks! I just worked through this data-set. I was getting an overall accuracy on the GPS of about 17.8cm with about 40 GCPs observed. I would be very interested in doing some Round Robin Testing with this dataset with you and others on the forum. The GCPs are rather difficult to pick out, but I did manage to centroid on quite a few of them. 

 

Which GCPs did you pick to orient the model? Could you share a mesh of the model? I'm going to generate my georeferenced mesh and will share it with a public drop-box link. It's a good set to work with with because of the good pick-up on the survey (46 points).