Just a simple question, in these two statements
E = cv::findEssentialMat(
points1, points2, focal, pp, cv::RANSAC, 0.999, 1.0, mask);
cv::recoverPose(E, points1, points2, R, t, focal, pp, mask);
Is the rotation and translation calculated from points1 to points2 or points2 to points1?
Thank you.
I tested it, it's the other way around if you apply rotation and translation to the camera 1 coordinates you get to camera 2 which makes sense considering the projection matrices are K[I|0] and K[R|t].
I'm working with it and I had the same issue. In the fucntion descrption it says that first you use previous points and then the actual points but I realised that it is worng. You need to introduce as first parameter the 'second points' and then 'previous points'. IN BOTH FUNCTIONS!
Without running a test I can't be 100% certain, but it should be camera 1 is (0,0,0) (0,0,0).
The R and t are the rotation and translation (in camera 2's coordinates) to get to camera 1.
Asked: 2016-12-20 20:40:46 -0600
Seen: 5,611 times
Last updated: Dec 12 '17
