Support Website Contact Support Blog

Orthometric Images, GCP's and why they aren't quite lining up

So i flew my first project with Pix4d the other day, and i used GCP’s generated by an rtk gps base/rover setup, and verified the GCP’s with a 2" total station to ensure absolute accuracy.  I used the GCP manager and painstakingly placed the GCP’s in all applicable pictures.  After processing and importing the Ortho from pix4d into Civil3d (great quality on the picture) i have noticed a difference of 0.3’ to 0.5’ on every single GCP, to make matters worse it’s not in a consistent direction, manually aligning the Othro image to the points in question yields even worse results, throwing the points im not holding out by close to 1’, Shouldn’t the image line up perfectly with the GCP’s provided? 

Thanks in advance for any tips/help/advice

Hey Kane,

If you have an RTK GCP points, it should be very accurate if you are clicking the right GCP’s. Is it possible that you are clicking the wrong GCP’s ? Do you have more than one GCP in a single image? That could be the source of discrepancy. Moreover make sure output coordinate system for an orthomosaic matches with the GCP’s coordinate system. What is the GCP accuracy in your Quality Report? Also are you inputting GCP’s in Basic Editor before Step 1 or after Step 1 using the Ray Cloud? 

Definately not clicking the wrong GCP’s as they are spaced out across the site in such a way that not more than one can be seen from any particular image.  I used the raycloud method of computing the gcp’s then re-optimized the model before sending it to the cloud for final processing.  I did output the ortho to the correct coordinate system (VA South Zone HARN) 

there is the gcp error report, it does look like point 10018 is quite far out

Hey Kane,

You are right point 10018 has a large RMS error, also point 10024 Y has a significant RMS Error as well. Maybe make a copy of your project then try disabling those two GCP’s and re-optimize. 

1 Like

So after eliminating the worst of the points and re optimising the project, then uploading the cloud and processing fully I am returning slightly better results with the remaining gcp’s, however in the areas near the eliminated control I am seeing a 2-3 foot difference in the ortho and the traditionally collected topography. There are some slightly blurry pictures and it was quite windy on  day I flew the mission so that might have something to do with the problem, but the drone didn’t seem to have a problem flying and it didn’t take very long. I flew at 125 feet with front and side overlap at 70%. Should i increase overlap? Fly lower/higher?

Dear Kane, 

Could you please create a ticket to our Technical Support with your request so we can investigate in a more efficient way the issue you are currently facing with your GCPs ?

Best,

Hey Kane, 

Increasing the overlap may definitely help with the overall  geo-location accuracy. Maybe try 70/85 overlap. I am not too sure though. What is the resolution that you are aiming for?