How can I calculate the rigid transformation [R|t] between two 3d triangles, but restricted to a given N degrees of freedom (for N = 1..6) ?

I know for N=6 I can get a least-squares solution via SVD of a certain matrix, but how can I integrate further constraints (fewer DOF) into the system?

  • 3
    $\begingroup$ Specifically what kind of constraints do you want to include? $\endgroup$ – Nathan Reed Apr 8 '17 at 5:29

Every 3D rotation has an associated space that can be generated by 3 basis vectors, which can be seen as the axis of the coordenate system of this space. These vectors are also the 3 columns of the transformation matrix. In order to decrease the dof of the rotation, you should provide one or more of these basis vectors and solve for the rest of the matrix. The same principle applies to the translation vector, you should provide constants for one or more distances of the translation. Of course, if you restrict the transformation too much you will end with an unsolvable exact system or a transformation that take the first triangle too far away from the second one, if you are using a non-linear solver.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.