I don't understand the parameters in the stereo case. It would be nice if you could elaborate on this a little. For example why are the parameters Camera.fx, Camera.fy etc. the same as in LEFT.P. However LEFT.K is not reflected in LEFT.P.
Isn't P = K[R|t], so for the left camera it should be basically the calibration matrix? Thus, LEFT.P = LEFT.K[I|0]. This is obviously not the case but for the general camera it is the case.
And this is only necessary if the images not already rectified isn't it?
I would be very glad if you could explain a little or refer me to papers/books/articles etc. which explains this.
https://github.com/UZ-SLAMLab/ORB_SLAM3/blob/ef9784101fbd28506b52f233315541ef8ba7af57/Examples/Stereo/EuRoC.yaml#L37
I don't understand the parameters in the stereo case. It would be nice if you could elaborate on this a little. For example why are the parameters
Camera.fx
,Camera.fy
etc. the same as inLEFT.P
. HoweverLEFT.K
is not reflected inLEFT.P
. Isn't P = K[R|t], so for the left camera it should be basically the calibration matrix? Thus,LEFT.P = LEFT.K[I|0]
. This is obviously not the case but for the general camera it is the case.And this is only necessary if the images not already rectified isn't it?
I would be very glad if you could explain a little or refer me to papers/books/articles etc. which explains this.