Stitching thermal

Hello,

Im using drones to protect wildlife in the Netherlands. Im trying to stich some thermal images in web ODM but its says cannot process data. The rgb images is working fine.

Are there special custom settings u need for stiching thermal images.

Thanks!

Lars

1 Like

Welcome!

Sorry for the trouble.

Can you start by giving us some more details about the thermal images, especially their resolution? They are often quite small and as such, are really difficult to find sufficient good tiepoints within.

It can certainly be done, but we usually have to push some settings.

If you’re able to share the dataset, that’s even more helpful.

1 Like

Thanks Saijin,

I have made a link with some data. The images are indeed small.

I hope u can help me out.

1 Like

What was your flightplan for these like?

It seems like overlap and sidelap are way too low, which can happen when the flightplan is designed for the primary camera on a mixed-camera sUAS that has a different resolution and focal length.

At this moment for rgb we use 10% overlap thats was recommended by our drone suplier because those meadows can be extremely long and u end up with 500+ images. Rgb images at 10% overlap will stich fine in web odm. When i did my research i have noticed that for stitching u use way higher overlap. Will that be the solution for better stiching those images?

We use a dji mavic 300 with zenmuse h20T.

I find it very hard to find information for the settings.

Thanks a lot already for helping me!


RGB images

1 Like

Maybe its interesting to tell what i want to do with the thermal images. We protect meadow bird on farm lands. we search for nest so the farmers knows where they are and not destroying them when working on the fields. Also we set some land underwater for food and shelter for those birds.

The farmer get compensation for the m2 wetland he has. So we need to measure the water part and also the soggy soil around the water part. At this moment we walk with gps around it and so doing the measurements. We are trying to find a way with thermal images to measure those wet and soggy part of a field.

1 Like

I would push overlap/sidelap as high as you can for the thermal… If you can, 65%+ in each axis at a minimum.

So i made a new thermal flight. I pushed overlap/sidelap to 65% and i flew 50meters high. its still cannot process the data.

I made a wetransfer link with all the flight images/data.

Are the images still not good?

1 Like

Is your overlap/sidelap percentage based off the main camera? That has a very, very different resolution and focal length to the thermal camera, so 65% overlap/sidelap calculation for the visible camera is nowhere near that high for the thermal camera.

These are looking something closer to 30% or so, which is yeah, not going to work.

1 Like

I use the mapping software from my DJI smart controller and change the overlap in camera setting only for the IR to 65%.

I succeeded to stitch the images from my previous post by change the web odm settings to field. But it’s is slightly off with the satellite layer underneath. Is it possible to get that little bit more accurate with some settings?

1 Like