Sequoia sun angle correction

Hi,

I haven’t found if it possible to use sun angle correction in Sequoia image radiometric calibration, found under “index calculator” tab in settings. We have license to use Pix4daq and also tested full version trial, but both list the radiometric correction “Camera, Sun irradiance and Sun angle” as red text. Should it be possible to use this correction for Sequoia camera image processing? And if yes, then why I cannot enable it?

Hi Teemu, 

For the moment the sun angle correction is available with these two drones: 

  • eBee from senseFly. Note that eMotion (senseFly’s software) needs to be used to extract the images in order to have the correct EXIF tags in the images. 
  • Disco Pro Ag from Parrot

The reason is that the geometry of the setup (drone, sunshine sensor, camera,…) needs to be known, which is the case only for the two setups above.

What setup are you using? 

We are using Airinov’s 3DR solo setup with Sequoia. So there is no tags for us. Do you have any info how these tags are defined and calculated?

I do not have any specific information as to how you could define these tags yourself. However, what I know is that the measurements to define the tags have to be very accurate, otherwise the error that is introduced can create more damage than value in the project. The measurements would be even trickier if you have a gimbal like on a 3DR Solo. 

I’d rather recommend to use the Camera and Sun Irradiance corrections. If possible with a radiometric calibration target on the ground for an additional reference.

Setup is fixed without gimbal etc. Then the sun angles are always quite low here in Finland which makes the sun angle problem worse. For example a slight wind can cause roll that makes noticeable quality loss in the final product due to sun highlights. Patterns are also generated depending on flight route directions and sun angle. Increasing overlap and flight height would fix it to some degree, but it is not possible to increase those anymore. It would be nice to try if there would be something to improve quality.

For NDVI its not that bad, because errors cancel out, but if you show reflectance images for audience, usually the first question is about these error patterns…

Would it be possible to ask someone what these sensor pose tags are? I’m sure pix4d programmers know these tag names, because you are reading them.

Thanks for explaining your use case. I’ve asked for more information and here is what I got. 

The EXIF tag is: “Xmp.Camera.IrradianceRelativeRotation” and has the following format: { Omega, Phi, Kappa } in degrees.

The angles are defined as described in this article: https://support.pix4d.com/hc/en-us/articles/202558969 

Below is an example of how the Sequoia is integrated in the eBee which is one of the supported drones mentioned in the first answer. In this case, the angles are marked as { 175, 0, 0 } in the Xmp.Camera.IrradianceRelativeRotation tag. Note how the Sequoia is oriented relatively to the sunshine sensor. The sunshine sensor is embedded in the top of the structure.

Here is a view from the side: 

and from the top: 

Here is more about the senseFly integration from their website: 
https://www.sensefly.com/fileadmin/user_upload/sensefly/documents/brochures/Sequoia_Specifications_2016_sensefly.pdf 

Note that the Sequoia and sunshine sensor are integrated in a fixed structure. They are placed very accurately and the angles will not change at any time during the flight. I would strongly recommend to create such a fixed structure for your integration as well, in order to reduce the error in the angles. 

An error of a few degrees in the measurement of the angles can cause large errors in the reflectance map, so please make sure that the angles are measured precisely. I can’t guarantee how well it will work, but I’m looking forward to have your feedback if you try it out!