GCP unchecked but will impact in camera reoptimising

[Phogi]

Hi Alexey,

I just tried to uncheck the GCPs to leave it as a checkpoint, but it seems the reoptimising will involve the "unchecked" GCPs in the camera model still. Would this be correct?

Could you tell me a bit about the formula used to reconstruct the GCPs with the camera alignment? As the unchecked GCPs how would it be weighted in the formula?

Thank you very much, any comments would be much appreciated!

Cheers,

[Alexey Pasumansky]

Hello Phogi,

For unchecked markers (check points) their source coordinates wouldn't be used during the optimization, however, the information about their projections on the source images would still be utilized.

[Phogi]

Hello Alexey,

Thanks for your super fast respond! However, as you can see in the screen snipping, the group 1(uncheck and optimised) vs group 3( deleted and optimised) have difference in the RMS values. Would this mean the value been used for reoptimising and dense reconstruction?

It would be of great help if you can give some hints of how the GCPs weight can impact on the dense reconstruction, as each GCP can impact how much area.

Thanks,
Cheers,

[Alexey Pasumansky]

Hello Phogi,

The projections of all markers (control points and check points) are utilized according to "marker projection accuracy" value. The coordinates of the check points do not matter - you should get similar optimization results even if you clear the coordinates for the check points, but keep their projections on the images.

The dense cloud is generated basing on the relative camera positions and estimated calibration parameters.

[Phogi]

Hi Alexey,

Thanks, but the projections are some hints of the data quality which related with the final DEM generated is it? If the camera RMS is higher, then the reoptimised RMS with unchecked GCP location will be different with "remove" the GCP at that location then, is this possible?

Thanks, for the GCP will it's weight impact on the nearest points reconstructed or there's a rule to georeference the point clouds later?\

Thanks,

[Phogi]

Hi Alexey,

So after some test runnings, I found the projections are not changing significantly to the point clouds. However, this raised me a question: how does the GCP values involved in the point cloud reconstruction? Does it just georeference the GCP coordinate to the sparse point clouds model, or it recalculate the points near the camera projections with the weight assigned? The first case looks like georeference the Lidar data with GCPs and rectifying, but if in the second case could we know how much area of the points will be impacted by the GCP while the other locations still impacted by the camera model?

Thanks,

[Alexey Pasumansky]

Hello Phogi,

The optimization process is refining the relative positions and orientation of the cameras (including the calibration parameters). 3D positions of the points in the sparse cloud are defined by their valid projections on the images, so 3D point locations depend on the estimated camera positions and orientation.

[Phogi]

Hello Alexey,

Thanks for the reply, apologise I missed the email notification! Since the 3D locations are reprojected, the GCP locations will be added in the reprojection with the weight, is there any other projections in the images will be forced to involve in near the GCPs to georeference the dense point cloud?

Thanks,