The initial image triplet used for the reconstruction method is represented by three meshes limited to a common covering area and a reference texture. Using the theory of homographic transforms, we extend the initial meshes to the entire reference texture.
 
  Consider two of the three original images of the 3D scene (e.g the fisrt view and the third one which plays the role of reference image of texture in this example). We can relate these views by a geometrical transformation: an homography H defined by a 3x3 matrix
we obtain:
 

Note that, in general the transform H only performs a 2D approximation between the two considered images, except in the particular case of single rotations using a unique camera (for the two different points of view). In this case, H is the expression of the 3D motion of the video camera from the position corresponding to the third point of view to the position corresponding to the first one.