CALIBRATION

Thanks to Ramani Duraiswami and

Zhengyou Zhang for the use of some
slides
 Shortcomings
Use precisely known 3D points

Shortcomings: Not flexible

very expensive to make such a calibration
apparatus.
 Zhangs approach
Use only one plane
Print a pattern on a paper
Attach the paper on a planar surface
Show the plane freely a few times to the camera
 Camera Model

x
M y
z
(u0 , v0 )

m
C
(R, t )

x
u u0 y
s v 0 v0 r1 r2 r3 t
z
1 0 0 1 R t


1
~
m A ~
M
 Plane projection
For convenience, assume the plane at z = 0.
x
u M
v y
(u0 , v0 )
z0
m
C

The relation between image points and

model points is then given by:
~ ~ with H Ar r t
sm HM 1 2
 What do we get from one image?
We can obtain two equations in 6
intermediate homogeneous parameters.

Given H, which is defined up to a scale factor,

And let H h1 h 2 h3 , we have
h1 h2 h3 Ar1 r2 t
This yields
T T 1
h1 A A h 2 0
T T 1 T T 1
h1 A A h1 h 2 A A h 2
 Linear Equations
B11 B12 B13
Let B A T A 1 B21 B22 B23

symmetric
B31 B32 B33
Define
b B11 B12 B22 B13 B23 B33
up to a scale factor
Rewrite
h1T A T A 1h 2 0
h1T A T A 1h1 hT2 A T A 1h 2
as linear equations:
Mb 0
 What do we get from 2 images?
If we impose = 0, which is usually the case
with modern cameras, we can solve all the
other camera intrinsic parameters.

How about more images?

Better! More constraints than unknowns.
 Solution
Show the plane under n different orientations
(n > 1)
Estimate the n homography matrices
(analytic solution followed by MLE)
Solve analytically the 6 intermediate
parameters (defined up to a scale factor)
Extract the five intrinsic parameters
Compute the extrinsic parameters
Refine all parameters with MLE