Support Website Contact Support Blog

Merging multispectral projects

I used an eBee with a Sequoia-sensor to cover a large archaeological site with multiple missions. Now I want to merge two or three missions into one project, while it works perfectly with the RGB-data, I ran into to some issues with the Multispectral-imagery. Somehow, the calibration is messed up and there is always a huge gap between the single missions.  The image below is showing two joined missions – one bigger northern one and smaller southern one. The border between the two missions is clearly to see. What is the right way to merge them? Thanks for your help!

 

Hi Matthias,

Do you see the pattern in the reflectance maps as well? If the flight has been taken on different days it is expected to have such differences in the index values.

Do you use a calibration target along with your project? If yes, please note that only one target is carried along with the merged project, hence the flights should have been carried consecutively so that a representative target should be applied.

Did you use MTP’s or GCPs for merging? If yes did you mark them in all the bands?

Could you post here the quality report file from both projects individually and the one from the merged one?

Thanks,

Ina

hello,

I tried to merge 3 blocks into one projects. After processing all 3 blocks individually through step1, I marked MTPs (with the same name) and then merged them and have processed step2. Everything seemed to be ok, but had some problems on step 3 with calibration target images - there were 12 calibration images in the “index calculation” tab - i suppose 4 images for each 3 blocks  (see the screenshot below - here only some of them visible), but the software could not find last 4 calibration images (for 3rd block. marked with red circle on the screenshot). Can I use one set of calibration images for this merged project? i.e. 4 images from one mission. Just for consideration: first mission was done on one day and the second and third one - on the next day with 2 hours interval inbetween.

And one more question - when merging multispectral projects, each from individual drone missions and individual sets of calibration target images for each mission, which calibration images should be used once merged and processed with step 3? Does the software automatically select corresponding calibration images for respective parts of the merged project? 

Actually, when projects are merged with duplicate camera parameters box ticked, it generates and seems to support multiple calibration targets, one for each project, but actually, when this happens, the band will not align. If there are 3 projects, this will act like 12 different cameras. Also, the reflectance maps will not be merged; you will get 12 maps. Hence this procedure would not be of your interest because you will get no extra benefit compared with processing projects individually.

So you will have to uncheck duplicate parameters. With merged multispectral images, we support only one calibration target. If from one flight to the other the sky coverage, time of the day (sun position), light conditions would change then it would be hard to find the image of the calibration target that would be representative for both of them. That being said relying on a picture of a calibration target from another flight in other conditions would compound errors and is not advisable. It is not implemented in the software. As your flights were taken at such a long time between them I would recommend not using the radiometric target anymore.  But if for some reasons you want to keep it then I suggest processing the flights individually and stitch the results in a third party software. 

However, if these conditions do not change then you have a representative set of images of the radiometric target then you can process them together.
For more details please see the paper: https://support.pix4d.com/hc/en-us/articles/204894705 

If you have 3 flights, then I would suggest you use the targets of the second flight (if all the flights were one after another).