Support Website Contact Support Blog

How to determine accuracy with no GCP

I do not have the equipment to collect GCPs.  I am using a Phantom 4 and have a client that would like rock stockpile volumetrics.  He is not concerned about Global accuracy, however, he stated they can accommodate a 100mm variance between the base of the stockpile and the top.

Without GCPs, how do I prove that the Relative accuracy is good and within the 100mm acceptable variance level.

Is there something in the Quality Report that can be used to prove the Relative accuracy or is this impossible without GCPs?

From your description of project, using high fidelity control is not needed, however would suggest taking some physical measurements of fixed objects (X,Y,Z) in and around the point of interest and applying that data the your project.

I have taken physical measurements of projects in the past as a way of ground-truthing relative accuracy and the results always have been very good, however, I have been doing so with the less than favorable method of using a physical tape measure.  Even though the tape measure method seems antiquated I see it as very accurate as there is no need to perform calibrations or know if the equipment is functioning, 10 inches on a tape is 10 inches.  I have always worried about the professionalism of using such a method of QA\QC as it appears to the client.

Confused, is the question to provide accurate data or something else?

Accurate data yes.  Were you referring to a method of taking physical measurements of fixed objects and inputing them into the Pix4D processing somehow?

I was referring to the fact of taking physical measurements of objects in the Area of Interest and comparing those physical measurements to those obtained by measuring in Pix4D and comparing the two.

Yes, note a few dimension in the area of interest and apply that data to processing chain.

https://support.pix4d.com/hc/en-us/articles/205360375-How-to-scale-a-project#gsc.tab=0

This helps validate your output and is cheap and easy assurance to your work flow.

I agree, that most often it seem unnecessary but if you see discrepancy in phase one processing it may saves time trouble shooting before phase 2-3 processing.

Just a suggestion , but best is high fidelity 3d reference points. 

1 Like

Thanks Gary.  I do not have the ability at present for high fidelity 3D reference points.  I was hoping there was a way to validate relative accuracy of the resulting data.  I’ll have a look at the link you just provided.

Can you give an example of, or explain you first point in you last response about applying data to the processing chain.  Are you referring to inputing that data into Pix4D processing somehow, ie high fidelity 3d points?  Or using some other method.

I really appreciate you help.

That article was very useful Gary, thank you.  So I can place GCP to use as Scale Constraints, know their exact measurements and than use this to correctly scale my project so I can verify its relative accuracy.  I believe this is exactly what I was looking for. 

I hope to be able to acquire a survey grade GPS at some point to use high fidelity 3d points but do not have that as an option at present but I think using Scale Constraints to verify my relative accuracy will suffice for my current project.

After I learned about using Scale Constraint I found this other forum here on the Pix4D site where another individual was attempting to do the exact type project I plan on completing.  Useful for anyone else looking for this information https://community.pix4d.com/t/5806-Capturing-the-pictures-to-my-first-project-How-do-I-get-it-right-from-start-#gsc.tab=0