computer vision - How do I combine transforms in the ceres solver? -
i have 2 parameters transformations input ceres cost function. both transforms combined, in order reproject points. both transforms given in form of rodrigues rotation vector, , translation vector.
my question is, how combine these 2 transforms within cost function (using ceres api's), in order reproject points?
have @ functions in ceres/rotation.h header file: http://ceres-solver.org/nnls_modeling.html#rotation-h
for example can convert rodrigues vector rotation matrix:
void angleaxistorotationmatrix<t>(t const *angle_axis, t *r)
with can build own 3x4 transformation matrix every transformation combining rotation , translation (you can use eigen http://eigen.tuxfamily.org/index.php?title=main_page that). matrix multiplication yields final transform (mind order).
Comments
Post a Comment