GCPs in large projects

I have to map an area with more than 200 m difference in elevation. I know I have to perform multiple flights over sections with somehow constant elevations and then merge all projects together.
I also know that there needs to be high overlap between each flight and at least 3 common GCPs in the overlapping areas.
If I mark each GCP in sufficient number of images (about 10 images) in each flight, when I merge them together in the end, each point will be marked in more than 20 images and I think it adds to my geolocation error instead of correcting it.
How I can find an optimum number of images that I have to mark, so each flight is georeferenced correctly but there are not too many for when I merge the project that it reduces my accuracy?
Thanks in advance.
Also, is there any suggestions for how many flights I should perform with this much of elevation change? one flight every 10m, 20m ?

Hi Raffaele,

First of all, we need to distinguish between relative and absolute accuracy (both valid for Pix4Dmapper):

  1. Relative accuracy: quantified by comparing individual features on the 3D model and the same features measured in the field (reality). For example, if a 3D model of a building shows windows two meters apart, and this is the same distance as on the actual building, the model has high relative accuracy. This can be improved locally (especially in areas with lower overlap or difficult image content) with the help of Manual Tie Points (MTPs). An MTP is a 3D point corresponding to a keypoint that is marked (clicked) by the user in the images. They are used for assessing and improving the relative accuracy.

  2. Absolute accuracy: defined as the difference between the location of features on the 3D model and their true position on the Earth. For instance, if the same 3D model of the building is not in the correct place on the map, it has low absolute accuracy, even if it has high relative accuracy. The absolute accuracy can be improved by using accurate geotags (RTK/PPK) or by applying a georeferencing process (GCPs).

In your case, you need to mark MTPs (not GCPs) in the common seams of your subprojects. Please read our extensive article dedicated to merging: How to merge projects.

You can also add GCPs to position your 3D model on Earth, but this should be performed in the final merged project. This way, you ensure that the georeferencing is global and consistent. The number of GCPs is based on a rule of thumb (arbitrary choice) and on the sigmoid curve of optimization (if adding more and more GCPs only yields minor improvements in accuracy, then you should use the amount of GCPs that showed the most significant decrease in error).

When marking GCPs or MTPs in the images, I would recommend marking them in all the images where they appear (this can be cumbersome though).

Regarding setting up a flight plan for mapping areas with elevation changes, I would recommend checking this article: Image acquisition plan for terrain with height variations.

Cheers,
Teodora

1 Like

Thanks dear Teodora,
I was doing it all wrong until now then! I would add many many GCPs and merge based on those points only since I thought the accuracy depends on the georeferenced points of GCPs. Honestly, measuring that many points with RTK on vegetated mountains was not very easy!!!
Your answer makes my life so much easier! I will not need that many GCPs then. Thanks as always.

Just 2 more questions though! One, after I merge based on MTPs, I should add a few GCPs, reoptimize, perform step 2 and 3. Is that correct?
Second, in vegetated areas without buildings or cars, how can I find MTPs? It is a bit hard to mark the same pixel of a stone or a tree! Thanks again.

Hi Raffaele,

I am glad I could be of help.

Regarding your first question, you first merge the subprojects (with processing step 1 already performed for each), then you mark the GCPs in the big project > reoptimize > process step 2 > process step 3.

To your second question: we are aware that the MTPs are difficult to find in vegetated areas. We have published an article dedicated to improving the outputs of vegetated regions, you might want to consider reading it: How to improve the outputs of dense vegetation areas? I strongly recommend searching for unvegetated landmarks in the seams of the subprojects, if possible.

If your area consists of 100% vegetation, then you can also try processing the entire dataset in one batch instead of dividing it into subprojects.

I wish I could help more, but this seems to be a challenging task you are facing.

Good luck!

Best,
Teodora

1 Like

Hi Teodora,
Thanks for your help.
It is outside of town and totally vegetated and on a mountainous area which has an elevation change of 170m


So I have to use multiple flights to consider the elevation change. I am considering 10 flights!
You are right, finding MTPs in this area is not easy! Thanks.

Can you use a capture app that supports terrain following? Map Pilot, which has terrain following, supports DJI drones, if that’s what you’re using. That would make your project much easier.

1 Like

Thanks @Andrew_Milanes. But I am using a Parrot Anafi and I heard Map pilot is not free.

Update to my question!
I just tried to merge my flights together without marking MTPs and surprisingly they were well put together ! I have not used any MTPs in my subprojects and I was going to add the GCPs in the final merged project. How can PIX4D Mapper put together my flights without knowing any common points? It did it by image overlap that existed between my projects? Is the result correct? If this happens, I can easily just import GCp sin my final merged project and live an easy life!

Hi Raffaele,

Indeed, the subprojects will be merged without the MTPs, because the software still finds the common features and matches based on texture and geometry. I assume that your subprojects have sufficient overlap. However, it happens very often that the matching is not optimal without the MTPs. The image content (e.g. vegetation, sandy areas), different lighting/meteorological conditions between flights can also play an important role. Additionally, you need to avoid calibrating your entire project into different blocks (to be verified in the quality report). I would still insert 3 MTPs in the seam areas, to make sure the relative accuracy and calibration are not affected.

Best,
Teodora

1 Like

I want to make a topographic Georeference levee of a 2000 hectare area, I have a Mavic 2 pro and mavic 2 platinum drone. I would like a precision around 10cm

  1. What height of flight do you recommend?
  2. How far away can I put my GCP
  3. Or can you suggest another method?

Hello,

  1. The flight height will depend on the drone camera parameter and the required GSD(ground sampling distance). To do the calculation, I would recommend you to try our GSD calculator.

  2. Regarding the number of GCPs, I would suggest you go through our support article on Number and distribution of GCPs.

Regards,