Large Sequoia+ project

[ljurje]

Hello everybody!

I have a question regarding processing multispectral images acquired with Parrot Sequoia+. I know there are some tutorials regarding this issue online, but  despite that I have a question.
So, I work on the project and we basically have more than 20 flights conducted with sequoia+ camera.
Before each flight we have taken the images of calibration panel, and during each flight we had a sunshine sensor. So basically I want to use both (panel and sunshine sensor) for each flight, and process all flights together to build a large structure.

However, I cant find the way to apply specific panel calibration to specific set of images. As far as I understand, Metashape uses single image of the panel for calibration of all images in the project?

Or maybe Metashape recognizes which panel should be used for which images based on the acquisition time? (i.e. Panel is imaged right before the flight etc)

Another question is - Do you recommend simultaneous processing of both RGB and MS images? I have a set of RGB images too for the whole area, but I am not sure what can I expect if I process them simultaneously?

I think these are some interesting questions and that I am not the only one who wants to know the answers!


Thank you!

[Alexey Pasumansky]

Hello ljurje,

Calibration panel data would be applied only to the images from the corresponding camera calibration group (in Camera Calibration dialog), so if you would like to have multiple flights be processed together, but have individual calibration panel for the reflectance calibration I can suggest to use the following approach:
- add each flight (together with the calibration panel images) to the separate chunk in the same workspace,
- use Merge Chunks option from the Workflow menu (to pre-processing of the data is needed),
- run reflectance calibration procedure and follow common processing pipeline.

With this approach of chunk merging you would have calibration groups formed automatically and you do not have to manually split the groups in the Camera Calibration dialog.

As for simultaneous processing of RGB sensor data together with multispectral Sequoia data, we do not recommend to do that. Maybe now something has been changed, but according to our experience RGB Sequoia sensor has very slow rolling-shutter and the images are severely distorted, if they are acquired while the camera is moving.

[ljurje]

Thank you Alexey!

Thank you for the step by step directions.
I have a few more questions.

Shall I leave only one set of panel images for each flight (by default sequoia takes 3 sets of calibration panel images), or Agisoft will use all three sets of images?
Also, can you provide a template for  CSV file for sequoia panel? I can find online csv template for only micasense red edge panels, not for sequoia panels.

And regarding RGB data, I was unclear there - I have Zenmuse X5 RGB images, not sequoia RGB images. Do you recommend simultaneous processing of hi-res RGB images and multi-spectral images, or it should be done separately?

Thank you for your thorough answer!
Brs