1 GCP with RTK drones

PIX4Dmapper
1

We are a survey firm, run a P4 RTK and use it primarily for topography and ortho tiles to aid in drafting. (we’ve never had any luck having the orthomosaic image to work).
What we’ve found is the RTK allows very good correlation between images and proper rotation but as would be expected it is shifted about 3’ from a State plane Grid localization and even more vertically. The easy fix it would seem is simply to use a single GPC to shift the array and move it vertically. As it seems the software requires a minimum of three GCPs my question is how to create MTPs, apply the same datum shift to them and use them as GCPs so as to prevent any scaling as in our experiences causes ghosting of the imagery on smaller sites.
Currently on small (>5ac) we process on straight through steps 1-3 to get the imagery, then process again with GCPs to get the contour data.
Of course any workflow other than my thinking is welcome!

Thank you.

2

Hi photchkiss,
I do not believe there is a way to understand what the error or as you describe datum shift is without having GCPs set. PIX4D always recommends using GCPs in projects to keep your accuracies (relative and absolute) as high as possible. Since the drone is acquiring positions in WGS84/GRS80 ellipsoid it is to be expected that you will have a significant vertical shift compared to your selected geoid. This will automatically correct for the proper vertical coordinate system if GCPs are used. If you are also seeing a shift in the horizontal due to using a modified stateplane then this will always correct for that as well.

I think the simplest solution is to continue using GCPs except you only need to process step 1. Then reoptimize, and continue on with steps 2 and 3.

As for ghosting, you may want to understand how to use the image editor to replace unwanted artifacts with clearer images.

3

Thank you Mike. I’ll try and find the image editor and read about it.

As far as GCPs. We generally do use 5 of them, but there are times we have an obvious photo point on a site with accurate x,y,z. As we are on a know localization (such as state plane) and we just want it primarily as a drafting aid, it would be handy to just use the single point as a GCP, then pick a few MTPs and apply the same x,y,z deltas or vectors to the MTPs. It would save a fair amount of time in not having to survey the GCPs. It certainly can be done externally by exporting the MTPs and transforming them and importing them as GCPs, but were there a way to do that within PIX4d Mapper it would stramline the workflow.

4

I guess I don’t fully understand. If you are already setting one GCP why not just set three? Then Mapper can utilize them to tie the project to your horizontal and vertical coordinate systems.

With that said, in theory you could use one GCP and calculate the x,y,z error. Then find the coordinates of an MTP. Apply that same offset, put it into a CSV file, and then load it into the GCP manager. Personally, I don’t recommend this as internal/external camera parameters, variations in lighting or image blur can often cause projects to distort and that same offset would not be constant throughout your project. You’re best to just set three GCPs with your GNSS equipment.

5

Can you export the RTK image geotags to a csv? If so, then import that csv into your desktop survey software to convert to grid coordinates and local datum. Then export converted points to csv for import into P4DMapper. This is the process I use with my Wingtra. You should get very accurate horizontal results with no GCPs. Vertical may be off by several tenths, however, without at least one GCP.

6

Yes, you can export the exif data. In the image properties editor you simply need to select the From EXIF option. Then you can manipulate the data and and upload using the From File option.

exif
exif1258×1060 267 KB