Support Website Contact Support Blog

Orthomosaic Accuracy

I’m working on a conceptual redesign of an intersection, and we usually use orthomosaics to set our alignments and LiDAR data that is available to us statewide for a DTM. The orthomosaics available to us statewide are pretty low quality, so I would like to start using orthomosaics created from Pix4D on all of our projects that require preliminary design. I flew a 63 acre area at 375’ with a Mavic Pro, and the orthomosaics created were very high quality. I used 5 GCPs spread around the edges of the project that were spaced 500’-800’ apart. My issue is that when I compare common points on the Pix4D generated orthomosaic to the orthomosaics available to us statewide there is a 3’-10’ discrepancy. Should I be trusting the Pix4D generated orthomosaic? Is there anything wrong with the process from which I generated the orthomosaic that I described above? I trust the accuracy of all GCPs used and they were all present in at least 15 images.

Could you share the quality report?

5 GCP’s in 63 acre (~25ha area) is too low in my opinion. Few days back I did a job 30ha job and used 43 GCP’s and 15 of the as checkpoints. That gave me enough information to verify the correctness of my otrho and DTM. 

Here is the quality report for the aforementioned project. I worked with Brad on the project and actually processed the data on my machine.

Can I asked what you used for GCP’s on your project? Setting 43 CP’s with targets would take quite a long time. 

You were flying Mavic Pro and it has rolling shutter. Try to run the project again with rolling shutter effect corrected

I used spray paint to mark GCP’s and RTK-GPS