Hello everyone,
I have a problem with absolute geolocation varriance for my master thesis project. I used DJI Phantom 4 Pro v2.0 drone and took photos. I used 6 GCP points which were measured with Trimble R8 (local coordinate system HTRS 96 TM) . When i look to EXIF data height of my photos they are ok (absolute and relative) and obtained 2D and 3D products in Site Scan have correct height ( HTRS 96 TM c.s.). In processing report absolute geolocation for X and Y is around 1 meter, and Z is -70 meters. How can i correct that 70 meters for Z and whats is cause of that. In attachment is my processing report.
Thank you!
MP
Solved! Go to Solution.
Hi @mpgeof24 ,
The "height error" you're referring to is from the "absolute geolocation variance" table from the processing report, this is the determined geolocation error of the input images. It's expected to see values in the range shown in your processing report when using images that have not been corrected using RTK or PPK. This does not affect the accuracy of your outputs because you used Ground Control Points, so there isn't anything that needs to be fixed in that regard.
How did you determine that the absolute height of your photos is correct? My suspicion is that it isn't.
Relative height should be quite accurate (measured above takeoff point). Select a photo that is near a GCP, look up its relative height and add it to the GCP elevation, does this match the absolute height recorded in the EXIF?
- Looking at your processing report, there's a relatively high geolocation error at your GCPs. If your GCPs were surveyed accurately, are clearly identifiable and were tagged/measured precisely, this error at the GCPs should be close to 0.
The GCP projection error column shows one value above 1 pixel. This may indicate that the GCPs were not tagged precisely and consistently, or not tagged in enough photos. I encourage you to review your GCP tags in each photo and ensure that they are precisely in the surveyed location. I also suggest adding a few tags per GCP to ensure you have at 6-8 photos marked per target. All GCP projection error values should be below 1 pixel, ideally near 0.5.
Regards,
Nico
Hi @mpgeof24 ,
The "height error" you're referring to is from the "absolute geolocation variance" table from the processing report, this is the determined geolocation error of the input images. It's expected to see values in the range shown in your processing report when using images that have not been corrected using RTK or PPK. This does not affect the accuracy of your outputs because you used Ground Control Points, so there isn't anything that needs to be fixed in that regard.
How did you determine that the absolute height of your photos is correct? My suspicion is that it isn't.
Relative height should be quite accurate (measured above takeoff point). Select a photo that is near a GCP, look up its relative height and add it to the GCP elevation, does this match the absolute height recorded in the EXIF?
- Looking at your processing report, there's a relatively high geolocation error at your GCPs. If your GCPs were surveyed accurately, are clearly identifiable and were tagged/measured precisely, this error at the GCPs should be close to 0.
The GCP projection error column shows one value above 1 pixel. This may indicate that the GCPs were not tagged precisely and consistently, or not tagged in enough photos. I encourage you to review your GCP tags in each photo and ensure that they are precisely in the surveyed location. I also suggest adding a few tags per GCP to ensure you have at 6-8 photos marked per target. All GCP projection error values should be below 1 pixel, ideally near 0.5.
Regards,
Nico