Hi,
We are testing our new equipment (Ebee X with RTK and a Sequoia+ with automatic radiometric calibration) and we got some strange results for the multispectral bands. I have played around with parameters but orthomosaics and reflectance maps always show stripes.
Our mission was a 40min flight over a 96ha corn field, with clear sky.
Any suggestions on wghat we may be doing wrong?
Hi, It seems like you flew towards and against the sun which can cause stripes. We recommend flying perpendicular to the sun. Sun angle correction usually works for such cases but it doesn’t work if the sky was overcast. Do you think it was overcast and not sunny? You can try processing with camera only calibration. Another reason of this can be low side overlap. As you can see from your quality report, some areas have really good keypoint matches and some do not. Do they seem correlating to your stripes?
Hi! Thanks for the reply!
Flying towards and against the sun does make sense… but It wasn’t exactly towards. The skype was completely clear, not clouds what so ever. As for side overlap, we had 60% lat and 80% long. And yes, I saw the quality report with the interlaced keypoint matches. I found that weird but can’t imagine why that would happen. They do somewhat relate to the stripes I see on the reflectance maps but do you think a higher overlap and flight lines directed perpendicular to the sun would resolve the issue?
This is my flight plan:
Thanks once again!
That always solves the issues users have with stripes. However, you can also send me the images, you can write us a support ticket if the data is confidential (mention my name along with a google drive link for the images). We have an internal tool for analyzing the sun angle, weather etc. I can generate some graphs and explain you in details why you got the stripes but i would need to full dataset for that.
Thanks! The data is free to use. I would love to have your help! I’m uploading all MSP and RGB images to the link below. The “therm” file goes along. It should be ready in an hour.
https://drive.google.com/drive/folders/1-lEmzTk1dH3TshYCVNMW0fM6yCZObwKI?usp=sharing
Hi Paulo,
These graphs might make it clear why the stripes are present.
Can you process with “camera only” correction and see if the stripes are removed? Side overlap/ keypoints also plays a role in the stripes. So from your next flight, I would recommend increasing the side overlap.
Thank you for the update! I processed with “camera only” and the results are a bit better but I can still see the stripes. I’m planning on flying again over the same area, on the same time, but with a different direction in relation to the sun, as you suggested. I’ll see if I get better results and post them here.
I do however have some questions before flying again:
One last thing, I used a 30 day trial for Pix4D Fields and processed the same images there. Results seem to be much better! However, I don’t know what parameters are used with Fields.
Hi Paulo, I will answer your questions below:
Should I fly perpendicular to the sun or sould I give some angle?
It is okay to have some angle, just don’t fly towards and against the sun (parallel)
Is flying interlaced lines a problem?
It is good to have uniform overlap, but if you mean double grid, you do not really need it. Double grid is good when you have 3D structures and need the facades.
What would be a good lat overlap? 80%?
Yes 80-85% frontal and 70-75% sidelap
For remote sensing purposes, should I use the orthomosaic tif files, reflectance tif files or the indices tif files?
You should use the reflectance tif files or index maps. The orthomosaic is color balanced and not radiometrically corrected.
In a future case scenario, where the sun if right over top of the plane, should I worry?
One think you need to keep in mind is it is recommended not to have shadows. That can interfere with the accurate index value/ reflectance calculations.
Pix4DFields works on a different algorithm altogether and we have seen it works with less overlap. So I think this can be an issue of less sidelap as both the software use the same radiometric module. Pix4DFields is great and you must have already seen how fast the processing is. Since it is a newer software, we are also adding new features and actively developing it.
Please let me know if I forgot something or misunderstood any question.
Dear Momtanu,
I think your plots for the analysis of flight conditions would be really helpful for a lot of people. Would it be possible to share some code for the generation or give some hinds with which output-files you are working?
Best regards,
Till
