I see a lot of posts and articles about number of GCP’s and I read through the blog post as well. It seems the sweet spot for vertical accuracy is 10-12 GCPs. However, these all seem to be smaller sites in the 20-40 acre range. We’re working on a 800 acre site right now and I am wondering if the same GCP’s to accuracy ratio applies or if I should increase the number due to the size. We’re really seeking vertical accuracy here. This is a 1.1cm GSD project with 8500 photos.
Hi Thomas, dialing in on the optimal number and distribution of ground control points is both an art and a science when you account for all of the factors that influence the optimal number of ground control points in a project, including, in no particular order:
- The geographical size of the project area. (Generally speaking, all other things being equal, if the size of a project grows, the total number of ground control points increases to maintain a sufficient density. Therefore, it’s reasonable to imagine that if 10-12 GCPs are sufficient for a 20-40 acre project, that you will need more ground control points for an 800-acre project to maintain the same density. However, in the real world, not all other things remain equal when the size of the project area increases. Therefore, decreasing the density of ground control points as the size of the project area grows, while it is not ideal, is reasonable.)
- How easily Pix4Dmapper can interpret the image content according to its texture, reflectivity, and stability, to name a few.
- How much image overlap there is.
- How accurate the initial image positions are.
- The processing options you apply.
- How accessible the project area is.
- How much time you are willing and able to dedicate to the process of measuring and then incorporating ground control points and how adept you are at doing both of those things.
- How accurate the results need to be for you to get the job done.
As a starting point, every Pix4Dmapper project should have at least five ground control points. How many more you collect in addition to the initial five will depend on the variables above. There are two schools of thought you can consider when it comes to designing the distribution of ground control points for a project.
One school of thought is to determine the optimal maximum distance between any location within the project area and the nearest ground control point. Unfortunately, there is no single distance that is appropriate for everyone due to the variables mentioned above. By far, the best way to determine the optimal distribution is by using your project data and then replicating an experiment that is similar to what the Nevada DOT did.
Another school of thought is to capture a certain number of ground control points for each mission that you fly. For example, if you plan to capture a project area with four overlapping grid missions and collect five GCPs for each mission, you would distribute as few as 13 and as many as 20. The reason that the total number of ground control points is not necessarily 20, in this case, is that two or more overlapping missions can take advantage of some of the same ground control points.
Please let me know if this doesn’t answer your question.
For such large projects having the right geoid supported can be important for good vertical accuracy, especially for corridor projects (e.g. roads,…). You could try Pix4Dmatic for that. Here are the supported geoids.
Out of interest (something in considering) do GCPs always have to be square or would a rectangular GCP do the same job, as long as one can discern the centre point? I want to make my own light weight GCPs and had the (brilliant) idea to use flat canvas boards (taking advantage of the already white background) but the large sizes that I could find were rectangular, although I’m sure I could find a canvas manufacturer to make them as desired. Is there a difference on the shape of the GCP, since round/ L-shaped GCPs seem to have been tested.
GCPs can be literally anything, as long as you can accurately measure and see a specific point on the images.
Feel free to check out the Ground control points: why are they important? blog post for more hints.
Hi, I’m doing a couple of lengthy projects lately, I’m from Argentina and need to use our local geoid model, will it ever be possible to add a geoid “.tif” file to pix4d mapper so I can achieve the most accurate results possible?
Hi @beckmannsantiago, we want to support more geoids in Pix4Dmatic in the future. Either by having more in the basis library of supported geoids or by allowing an import of geoid files. The exact way we’re going to do it still needs to be defined. Which one exactly would you like to use? This will help prepare the moment when we’ll add more support. Thank you.
The one I use the most is the Ar16 Geoid, it can be found in the national’s geographic institute web site: https://www.ign.gob.ar/
I am more interested in using it on Pix4Dmapper though, there should be a way to import the geoid file.
Thanks for the info, I’ve added it to the Pix4Dmatic and Pix4Dmapper suggestions.