% using the n-point homography estimation functions:
%
cp = [ 0 0 ; 0 1 ; 1 0 ; 1 1 ]; % any number of points > 3 is OK
lp = [ 10 10 ; 10 11 ; 11 10 ; 11 11 ]; % these correspond to cp

% the following lines
%   first append a third coordinate of 1 to each point
%   then take the transpose
% to put the data into the format that homography2d requires
cpReformatted = [ cp, ones(4, 1) ]';
lpReformatted = [ lp, ones(4, 1) ]';

% this returns the 3x3 homography, H
H = homography2d( cpReformatted, lpReformatted );

% this creates a Matlab transform structure named tform
% Here, we need to use the transpose of H in order to
% meet the expectations of Matlab's maketform function
tform = maketform( 'projective', H' );

% Here is the output from the above run:


>> H

H =

    0.5774   -0.0000    5.7735
   -0.0000    0.5774    5.7735
   -0.0000   -0.0000    0.5774

>> H / H(2,2)

ans =

    1.0000   -0.0000   10.0000
   -0.0000    1.0000   10.0000
   -0.0000   -0.0000    1.0000