I took pictures of the facade of the building under several points of view. Then I use bundler to restore the building point 3D cloud. Then I use the RANSAC method to find the facade plane in the point cloud.
This gives me parameters for the plane and its normal surface.
From the bundle, I also get camera rotation matrices, which I use to calculate camera viewing directions.
Now, given that I have a normal plane and a camera viewing direction, I would like to use this to correct images, that is, as if the images shot with the camera viewing direction were the same as the normal plane.
How can I calculate the projective transformation matrix from these two vectors?
source share