Support Website Contact Support Blog

differences in point cloud elevation from separate flights

I have processed a set of images from two separate flights executed in short succession. One is ta grid of a large area, and another was the detail of a material pile that I shot using an orbiting method. 

I did not set down GCPs (i believe GCPs do not help with volumetric calculations?).

There is a marked difference in the elevation of the detail area and the total area, as shown in the images attached.

My questions:

  1. What is causing this discrepancy?
  2. Should I fix it in the image capture
    portion of my workflow or in processing?
  3. How can I correct this?
  4. What are the red bars demarcating the difference between the identical features? 

Hi Taylor, 

Thanks for posting screenshots, it helps to understand what happens. 

When you process images from two different flights in the same project it can happen that the software is not able to match the images between the two flights on its own.

There can be various reasons such as a different illumination or shadows between the two flights, or if the Ground Sampling Distance (GSD) of one flight is more than twice the GSD of the other flight it can get difficult to match keypoints between the two flights, see this article: https://support.pix4d.com/hc/en-us/articles/202558979

In your case I would do the merging procedure described in this article: https://support.pix4d.com/hc/en-us/articles/202558529

Summary: you have to process step 1. Initial Processing separately for each flight, then you add Manual Tie Points (MTPs) at the same spots in both project (reoptimize each of them separately) and then you merge the projects. Once merged you can process steps 2 and 3 and compute the volumes. Adding the MTPs gives additional information to the software to match the images from the two flights. 

Note that Ground Control Points (GCPs) can help improve the accuracy of the volume measurement. GCPs are usually used as reference for the absolute positioning (e.g. on a coordinate system) of a project. However, they also provide reference points for the relative reconstruction of a project, which can be improved with that information. If you decide to use GCPs, I would recommend to place them close to the stockpile if possible.  

Another solution would be to use scale constraints, here I would also recommend to measure something close to the stockpile so that the influence of the scale constraint is stronger. More here: https://support.pix4d.com/hc/en-us/articles/205360375

As for the red bars you mentioned: this is the software not knowing where the top of the stockpile is, because of the two layers. Usually, the volume of the stockpile below the top pixel layer is filled by red to show what volume is measured. 

Let us know how it went!