I give the following example to illustrate my question: suppose I have a camera intrinsics matrix that follows the pin-hole camera model, and its intrinsics matrix has the following structure:
ori_intrinsics= [[fx, 0, cx,],
[0, fy, cy],
[0, 0, 1]]
I also know the image it can generate of size [ori_width, ori_height]
. Then now I crop the image in the image position [ori_left, ori_top, ori_right, ori_bottom]
. Now I have a new image. Then my question is as follows: image now I have a similar camera(the same pin-hole camera) and it will generate the same new image without any cropping operation as I did before. What will be the camera intrinsics for this camera?
Cropping an image does not affect focal length, so no change to fx, fy
. The only thing that will change is cx and cy, and this needs to be adjusted based on how much you have cropped from top-left.
If you crop by L, T
, then:
c_newx = cx - L
andc_newy = cy - T