Document the order of the coefficients in rs2_intrinsics::coeffs #7061
Conversation
foohyfooh
force-pushed
the
documentation/rs2_intrinsics_coeffs_order
branch
from
891a725
to
e2ca808
Compare
There was a problem hiding this comment.
@foohyfooh , thanks for the contribution.
The order is correct, however it applies only to the distortion model that utilizes radial and tangential coefficients.
In case of FishEye, for example a different (Theta-distortion) model is used with four coefficients K1..K4.
I suggest to adjust the wording to reflect that the coefficients [1..5] can be read [k1,k2,p1,p2,k3] where applicable, but other interpretations are possible as well.
foohyfooh
force-pushed
the
documentation/rs2_intrinsics_coeffs_order
branch
from
e2ca808
to
4676e23
Compare
|
From what I saw about Kannala-Brandt distortion model, there appears to be an eight parameter model with the parameters being [ fx, fy, cx, cy, k1, k2, k3, k4]. So would I be correct in assuming that the four coefficients in the array are [k1, k2, k3, k4, -]? |
|
Correct, for Fisheye those are mapped to K1..K4 ( coeffs[4] is undefined) |
There was a problem hiding this comment.
@foohyfooh , sorry for the delay and thank you for the contribution.
Much appreciated!
I could not find the order of the distortion coefficients in the Realsense documentation so I am adding them to it. I am assuming that they follow the order that is used in OpenCV for its distortion coefficients.
Feel free to correct me if this order is wrong.