Hi Pix4D Technical Support,
I am a user of Pix4DMapper, and have a question about technical details of the georeferencing method used in Pix4DMapper.
I understand that there are two different georeferencing methods: Non-rigid registration and Rigid registration when photos are tagged with GNSS coordinates.
- Non-Rigid Registration (adaptive registration) [SfM]
- Pose prior (GPS/Odometry) can be used as a non rigid prior during SfM Bundle Adjustment phase.
- Rigid Registration [SfM]
- Once you have performed SfM on your scene, you can adjust your scene coordinate system to fit your GPS datum.
Could you let me know which method is used in Pix4DMapper, please?
I believe the answer is that Mapper can do both. If using the Standard calibration method, Mapper will rely on image content to calibrate the scene. If the images do not have geolocation, it will still create the model and GCPs can be later incorporated for georeferencing.
However, if the images have geolocation and orientation data (particularly RTK/PPK positioning), then you can use the Accurate Geolocation and Orientation method for calibration. In this case, Mapper will place more weight on the geolocation of the images when building the scene. In this case, georeferencing is done at the same time as calibration.
I hope this answers your question.
Let me clarify my question further. In case that all images have geolocations from RTK-GNSS, and there are NO GCP measurements at all, then which approach is used for georeferencing in the mapper, the rigid registration (independent of SfM) or the non-rigid registration where the prior geolocations from images are used in SfM bundle Adjustment phase? If the mapper supports both, then where can I find the corresponding option in the menu of the mapper?
Thank you very much in adavance.