c++ - undistortPoints, findEssentialMat, recoverPose: What is the relation between their arguments? -

in hope broader audience, repost question here asked on answers.opencv.org well.

tl;dr: relation should hold between arguments passed undistortpoints, findessentialmat , recoverpose?

i have code following in program, k , dist_coefficients being camera intrinsics , imgpts. matching feature points 2 images.

     mat mask; // inlier mask      undistortpoints(imgpts1, imgpts1, k, dist_coefficients, noarray(), k);      undistortpoints(imgpts2, imgpts2, k, dist_coefficients, noarray(), k);       mat e = findessentialmat(imgpts1, imgpts2, 1, point2d(0,0), ransac, 0.999, 3, mask);      correctmatches(e, imgpts1, imgpts2, imgpts1, imgpts2);      recoverpose(e, imgpts1, imgpts2, r, t, 1.0, point2d(0,0), mask); 

i undistort points before finding essential matrix. doc states 1 can pass new camera matrix last argument. when omitted, points in normalized coordinates (between -1 , 1). in case, expect pass 1 focal length , (0,0) principal point findessentialmat, points normalized. think way:

1. possibility 1 (normalize coordinates)

    mat mask; // inlier mask  undistortpoints(imgpts1, imgpts1, k, dist_coefficients);  undistortpoints(imgpts2, imgpts2, k, dist_coefficients);  mat e = findessentialmat(imgpts1, imgpts2, 1.0, point2d(0,0), ransac, 0.999, 3, mask);  correctmatches(e, imgpts1, imgpts2, imgpts1, imgpts2);  recoverpose(e, imgpts1, imgpts2, r, t, 1.0, point2d(0,0), mask); 

2. possibility 2 (do not normalize coordinates)

    mat mask; // inlier mask  undistortpoints(imgpts1, imgpts1, k, dist_coefficients, noarray(), k);  undistortpoints(imgpts2, imgpts2, k, dist_coefficients, noarray(), k);  double focal = k.at<double>(0,0);  point2d principalpoint(k.at<double>(0,2), k.at<double>(1,2));  mat e = findessentialmat(imgpts1, imgpts2, focal, principalpoint, ransac, 0.999, 3, mask);  correctmatches(e, imgpts1, imgpts2, imgpts1, imgpts2);  recoverpose(e, imgpts1, imgpts2, r, t, focal, principalpoint, mask);   

however, have found, reasonable results when tell undistortpoints old camera matrix shall still valid (i guess in case distortion removed) , pass arguments findessentialmat if points normalized, not.

is bug, insufficient documentation or user error?

update

it might correctedmatches should called (non-normalised) image/pixel coordinates , fundamental matrix, not e, may mistake in computation. can obtained f = k^-t * e * k^-1

as turns out, data seemingly off. using manually labelled correspondences determined possibility 1 , 2 indeed correct ones, 1 expect.