 
    
    
         
The perspective projection from Euclidean 3-space to an image is represented as


 is a 
homogeneous
 is a 
homogeneous  camera projection matrix (11 dof) with the 
decomposition
 camera projection matrix (11 dof) with the 
decomposition

The algorithm for camera calibration has two parts:
 from a set of points with known
3D positions
and their measured image positions.
 from a set of points with known
3D positions
and their measured image positions.
 into
 into  and
 
and  via the
 via the  decomposition.
 decomposition.
 1. Compute the matrix  : Use correspondences between
3D points
: Use correspondences between
3D points  and their 2D images
 and their 2D images  to determine the matrix
to determine the matrix  .
.
 .
.

multiplying out

 ) correspondences,
a linear solution can be obtained for
) correspondences,
a linear solution can be obtained for  from the
set of 2n linear simultaneous equations (cf computation
of a projective transformation):
 from the
set of 2n linear simultaneous equations (cf computation
of a projective transformation):  ,
where
,
where  is the 12-vector representation of the projection
matrix
 is the 12-vector representation of the projection
matrix  , and
, and  is a
 is a  matrix.
The solution is the eigenvector with least eigenvalue of
 matrix.
The solution is the eigenvector with least eigenvalue of  .
.

Example - Calibration Object
         
            
Determine accurate corner positions by
The final error between measured and projected points is typically less than 0.02 pixels.
 2. Decompose  into
 into  and
 
and  :
:
 The first  submatrix,
 submatrix,  , of
, of  is the
product (
 is the
product ( ) of an upper triangular and rotation matrix.
) of an upper triangular and rotation matrix.
 into
 into  using 
the
 using 
the  matrix decomposition. This determines
 matrix decomposition. This determines
 and
 and  .
.

Note, this procedure produces a matrix with an extra parameter k

with  , and
, and  the angle between the image axes.
 the angle between the image axes.
 
 
    
   