Point Cloud Shifted?

I’m trying to figure out why part of the point cloud has shifted down in one area. I have a couple of ideas why but do not know how to fix this issue:

I had to remove one GCP (near the area of the shift) because it had been removed prior to the flight by an unknown source. I currently have four GCP’s that I thought would be enough to build the initial cloud. 

The main area of focus is in the center, so the surrounding edges won’t matter since I will be cutting the Ortho to the area of interest. If I continue to run step 2 and 3 with this anomaly will that effect accuracies or quality of the Ortho? Here is a screenshot:

 

 

Any advice would be appreciated of how to move forward with the processing!

Hi Matt,

 

Could you post the quality report here so I could take a look?

 

Cheers,

Ina

It won’t let me upload the entire report so here are some screenshots

Thank you for the screenshots.

I believe that the cause of this issue is the images from take-off and landing:

 

Could you create a new project without the images within the red square above and process again?

 

Thanks,

Ina

 

 

Hi,

 

I disabled those images and re ran the project and it seemed to fix the problem- but I still don’t know what the exact cause of why the point cloud was misaligned in that area. I looked at the geolocation and all images were consistent in elevation.

 

Before disabling the images, I started over a second time with the just the four GCPs and the initial report looked fine. I then ran the point cloud and then it misaligned again- is it possible there is a bug causing this?

I would say that having overlap is more critical than the geolocation. Since the overlap between these two lines is not satisfactory the reconstruction of the scene is expected to be poor. Moreover, these images may be at a height difference that would result in a GSD twice higher than the project.

To correct this part, you will need to add GCP’s or MTP’s precisely in the part of the project. However please note that the accuracy on the edge of the project is always expected to be lower since the scene is seen in fewer images (less overlap). Hence we always recommend that the flight plan to be slightly bigger than the area of the interest to assure the best accuracy of the project.

As per the second part, if I correctly understood, you had added the GCP and then ran step 2?

After adding the GCP’s have you reoptimized?

Additionally, could you tell me which version are you currently using?

It would be handy if you could send via google drive or we transfer the quality report, the log file, and the p4d file so that I better understand.

Cheers,

Ina

 

Hello again,

I’m reaching out because I’m experiencing the same issues again with a shift in the point cloud (same project). I have checked with the pilots and made sure that it was nothing on our end that we did. We have the same GCPs, the geolocation file is consistent with the elevations, and the original photos look fine. 

The screenshot below illustrates the output after running a Full Initial. I’m only experiencing this shift after running a Full Initial, and it’s happening in just this area, which is located by a dredging pit. I ran it on rapid before and did not  have any issues. When I click on the camera location it is giving me an elevation of 161.46 ft, and the initial and orientation is giving me a value of 416.55ft. Any advice or suggestions on how to fix this would be great, I don’t understand why I’m having a hard time with just this project. It’s odd that all previous times have been fine, just only sometimes this happens with the same project…

Hi Matt,

 

Could you please send me the quality report so I could have a look?

After step 1 have you marked the GCPs and re-optimized?

Do you use the latest stable version of the software?


Additionally, to ensure that we are on the same page I will make a summary of my understanding:

- the takeoff and landing images have caused the initial shift

- after disabling those images and adding the GCP’s the project aligned correctly

- after re-running step 1 are the marks of the GCP’s in the project?

 

Thanks,

Ina