How to calculate a rotation matrix between two 3d points

Hi everyone!

I have two vectors that represent one point with respect to two different reference systems, eg, p0=[x0, y0, z0] and p1=[x1, y1, z1]; I need to know wich is the rotation matrix that transform the vector p1 to the vector p0. I absolutely don't know the angle rotation, neither the axis around wich the rotation is carried out.

I've tried to use 'vrrotvec' function and then 'vrrotvec2mat' to convert rotation from axis-angle to matrix representation; in theory, if I use this two functions to calculate the rotation matrix R between p1 and p0, when I compute R*p1 I should obtain p0, but the outcome is a vector different from p0.

I hope I've been clear!

Do you have any suggestion?

Thank you!!

Answer 1

[Jos (10584)]

You can used dot and cross products to get the rotation matrix:

% two random 3D vectors
p0 = randi(10,3,1)
p1 = randi(10,3,1)

% calculate cross and dot products
C = cross(p0, p1) ; 
D = dot(p0, p1) ;
NP0 = norm(p0) ; % used for scaling
if ~all(C==0) % check for colinearity    
    Z = [0 -C(3) C(2); C(3) 0 -C(1); -C(2) C(1) 0] ; 
    R = (eye(3) + Z + Z^2 * (1-D)/(norm(C)^2)) / NP0^2 ; % rotation matrix
else
    R = sign(D) * (norm(p1) / NP0) ; % orientation and scaling
end
% R is the rotation matrix from p0 to p1, so that (except for round-off errors) ...
R * p0      % ... equals p1 
inv(R) * p1 % ... equals p0

[alessiadele]

Thank you very very much! it works perfectly!! :D

[Zhe Chen]

This is an useful answer. But please be aware that for two given vectors p0 and p1, the rotation axis and rotation angle is not determinant.

In this answer, the rotation axis that is used is the axis that is perpendicular to both the vectors p0 and p1.

[Lucien Robinault]

Hello,

First of all, thanks a lot for this.

Would you have any references on this way of doing by any chances? It is definitly not the way it is explained on all the ressources on rotation matrix (which I couldn't manage to get working anyway), so I would really like to finally understand the process involved in finding the rotation matrix.

Thanks a lot

[Kevin Hughes]

Thanks this was very helpfull