Units of image points for solvePnP
Hi, I am new to this subject, so please forgive the silly question. To use solvePnP, are the positions of image points needed in pixel coordinates or mm? I.e. my 2D images are x ray images. Do I need to consider image spacing?
Thanks
it's all in pixels / voxels, not mm(opencv does not know anything about your resolution/dpi here.
(just curious, - how do you get your 3d correspondences from the x-ray images?)
Currently, I define correspondences manually. Do you have an idea for automating this?
depends on what you're doing. in my simple case, it was for faces, and i could easily use a premodelled 3d head, and dlib landmarks for the 2d points
For my understanding: The imagePoints are given in pixels. I have the objectPoints in 3d coordinate system of a tracking system in mm. Now: the pose of the object in which coordinate system??