dim.kyr
 lvl.2
Switzerland
Offline
|
Hello everyone,
I am writing here my question as I am new and I am not allowed to open a new topic. However, my question is related only to the new firmware attribute "Added Oblique GSD in DJI Pilot."
It is very useful for all of those who collect data having tilted camera and not facing down (90deg).
However, my main questions are about the respective calculation of it.
Please let me explain my thoughts which can help you understand my question and might be useful also for other users:
In case of face-down camera, situation is like the following image:
Thus, we can use the two equations that you also mention on enterprise-insights.dji.com/blog/ground-sample-distance
GSDh= flight height x sensor height / focal length x image height;
GSDw= flight height x sensor width / focal length x image width
The two GSD values (h and w) I assume are applied for the whole pixel of image.
However, in case of tilted camera situation is very different:
Above image represents the situation of a camera capturing data with an angle different than 90 deg, viewing it from the side.
As it is obvious, assuming that there are not objects in the field to block the view, area closer to the image basement will have much better (small) GSD than the area at the top of the image.
Thus, we can have lets say a best and worst GSD scenario for the image about the min (distance OA) and max distance (distance OC) to the ground.
There is also the Average measured distance (AMD) which is the one which cameras with distance sensor measure.
Based on all the above I assume that we can use the respective tilted distance (OA or OB or OC) to the
" GSDh= flight height x sensor height / focal length x image height " equation and calculate the respective GSDh value for the respective part of the image.
Respectively, we can do the same for the width direction of image. It is important to mention that in cases of tilted camera with angle different than 90deg, image shape is a trapezoid with small basement close to the camera and bigger one at the top (see following image).
In the controller at the moment, there are two presented GSD values, one for nadir capture and one for tilted ones.
Based on all the above, I think it is very important to:
-Clarify in which case (distance/area of captured image) the tilted GSD is referred to. Is it referred to the worse scenario which is the point C, is it referred to the middle of the camera (point B). or where?
- Distinguish the GSDh and GSDw as they can be very different between themselves in cases of tilted images. I know that generally the worst of the two values is used but in some cases it is useful the user to have both of them in order to be able to define better also the front and side overlap.
Moreover, about the overlap definition, how the software define the distance between two consecutive images?
Assuming we have two consecutive images with side overlap as they are shown in the following image:
First image is the red one and then is coming the green one having an X % of side overlap. How the overlap is calculated?
In practice there are two ways:
-Per basement: so distance PC is the X % of the distance EC
-Per area: so the area PKBC is the X % of the area ABCE.
The two above mentioned ways give different results depends also on the camera angle. Thus, the methodology of images overlap calculation, especially for side overlap, is very useful to be known to some users with engineering background.
I know that for some users all the above information might be too much details but for some engineering companies who cares a lot for the final accuracy and try to optimize the data collection workflow it is very important.
Thanks once again for the amazing job you do and the continuous support. 
|
|
Administrator
Captain
DJI team
