I extracted the sensitivity of a multitude of flights we conducted with an eBee X and one Parrot Sequoia sensor. The results can be seen in the attached image. I’m a bit worried about the large range of sensitivities (that are used as absolute values for correction of final reflection?).
Might this be a hint for some problems in our acquisition process that lead to substantial error in results?
Example for sensitivity from log file:
“[2021.10.12 15:42:54][ 13%RAM][ 83%CPU][Info]: Mean corrected pixel value 0.468, std dev 0.0134712, sensitivity [ 0.86191 ]”
Our team recommends reading the following article.
Thanks for the article but what does it mean in the context of my question? Do you recommend not to use a calibration target? Is it true, that the plotted sensitivity should not vary in this rang or do I miss something during processing?
Sorry the article was for the other questions you asked on the support ticket…
I have forwarded you question to our team, they will get back to you.
First of all, it is very important to verify that the illumination conditions during the capture of the reflectance targets are representative of the illumination during the flight. Make sure that there are no shadows on the reflectance target nor on the sunshine sensor, and neither reflections onto the reflectance target from adjacent objects or yourself, and that the general illumination didn’t change (for example due to clouds forming). Make sure to hold the camera horizontal, and align the target horizontally and facing North.
Secondly, if you are using PIX4Dmapper, remember that as mentioned in the knowledge base, the sun angle correction assumes clear sky. If you have overcast conditions, please select “Camera and Sun Irradiance”. In PIX4Dmapper, selecting sun angle corrections in cloudy conditions leads to biases. In PIX4Dfields, you select the weather conditions in the settings.
For further investigations, I would suggest to plot the reflectance target derived sensitivity as a function of the weather condition. For those cases where the sensitivity is quite different across bands for the same flight, I would suggest to check if you selected images from the same capture for the different bands. If you mix two different captures that were acquired in different illumination conditions, you get biases. In general, I would suggest to always note down the weather conditions during a flight and take a few pictures of the sky with your mobile phone to visualise the in-situ conditions to refer to them later.
Thanks Klaus for your reply.
Thanks a lot for your detailed reply. First of al I have to admit, that we must be more thoroughly during the capture of the calibration images. It would be great, if you can additional answer following questions:
- When the sensor additionally measures sun irradiance, in my understanding sensitivities should be stable across illumination conditions. Am I wrong?
- When sun angle correction is activated, the irradiance of the calibration image is corrected, too?
- It would be great to analyze the measured irradiance for the calibration images for some analyses, but these are not exported to the sun_angle.csv. Is there any other way to get them.
When using the sunshine sensor, the sensitivities should be stable across illumination conditions if (1) the illumination conditions of the sunshine sensor are the same as the illumination conditions of the reflectance target, (2) the proper corrections are done, i.e. the right correction type for the conditions is chosen, and (3) the orientation of the camera is correct.
When sun angle correction is enabled, it is also applied to the reflectance target image. However, since photogrammetry doesn’t apply to a single image as the one of the reflectance target, we need to make some assumptions regarding the orientation of the camera, therefore the sun angle correction will be wrong if the camera is not oriented correctly.
Thanks a lot for your detailed technical insight and sorry for responding so late. Just on more small request. What do you mean by “the orientation of the camera is correct” in this context?
You should orient the camera as explained in my first reply: