Starting from:
$35

$29

Homework 07 Solution


Note the following class changes:

Prof. Gennert will be away the week of Nov 4. To make up for lost class time, we will meet as

follows:


Mon 28
Oct.
4:30-7:20pm with breaks. HW #7 due.

Wed 30
Oct.
4:30-5:50pm as usual.

Mon 4 Nov.
No class. HW #8 due.

Wed 6 Nov.
No class.

Mon 11
Nov.
4:30-7:20pm with breaks. HW #9 due.

Wed 13
Nov.
Resume usual schedule.




⃗⃗

















1.  Singular Value Decomposition (4 pts): This is an important tool that allows us to solve

systems of equations
    ⃗ ≈ 0,
|  ⃗|
= 1. This is useful in camera calibration, where we get

such a system of equations and want a non-zero parameter vector   ⃗. See Szeliski for an


introduction to SVD.























You can think of the SVD as follows:
  ⃗ in the
direction, scales it by



1.  It takes the component of input


and











1
direction.

1





1



outputs it in the




















,















2.  Repeat for all

, and   .












3.  The output vector is the sum of all the contributions









= [  |

] =
1


[1 |
1

0
]=[2  0],   =[

] =
1
[
1
1
]









], =[1


1








a.  Suppose that 2×2 matrix M has singular value decomposition









1
2





1 −1
0


0  1 1





















2





1  −1




√2







1√2









2








Without computing M, predict the value of     ⃗ for   ⃗ = [1] and   ⃗ = [−1] based on





















1


  ⃗, and finally, so does  ⃗. Now compute M and use it to verify    [1] and
the SVD properties. Hint: Note that in both cases
  ⃗ has a simple form, so does

1























[−1].






















b.  Let matrix   have singular value decomposition    =. Show that matrix

has SVD=   2
.














    2. Camera Calibration (10 pts): One restriction in camera calibration is that the world points
chosen must not lie in a single plane, that is, they cannot be   ⃗co-,planar,1≤ ≤otherwise calibration

will fail. To see this, suppose that there are K world points. We know that we
a single plane represented by   ̃ = (  ,   ,  ,   ), defined by   ̃
∙   ̃ = 0 (Szeliski, eqn.
2.7).
need at least 6 points for calibration,   ≥ 6. Consider what happens if all world points lie in



Image Plane


Plane containing
































·

1






















1
·






































































·












·

































⃗⃗




















The calibration equation is given by     ⃗⃗⃗ = 0, written out as


















1
0
0
0



0
























0










1  1
1  1
1  1
1

1

1

1


1







1















2

0

0


0

1
1























]


0


0





1



1

1  1
⋮1  1
1  1
1  [






0


















12

























⃗⃗



⃗⃗





  ⃗⃗⃗ that we
Ideally, there should be a single vector   ⃗⃗⃗ such that     ⃗⃗⃗ = 0 (or ≈ 0). This is the

hope to find through the singular value decomposition of   . The key concept here is the rank
columns. For this
  ⃗⃗⃗
to exist and be unique,

must have rank = 11 (or 12 with a very small
of a matrix, which is the number of independent rows, also the number of independent

value for the smallest singular value    12. Show that if all world points are co-planar, then
    ⃗⃗⃗cannot= 0⃗⃗have rank greater than 9 by finding 3 independent non-zero vectors   ⃗⃗⃗ such that
. (These independent vectors establish that the nullspace of    has rank at least 3;
hence    ⃗⃗⃗’s=rank0⃗⃗ cannot exceed 12-3=9.) In this case, it is not possible to find a unique   ⃗⃗⃗ such
that    and the calibration  ⃗⃗⃗ procedure fails.
Hint: The non  ̃-zero∙  ̃ =vectors0 are mostly 0s. Use the fact that if all world points are co-planar, then .



3.  Focus of Expansion (8 pts): Suppose that the viewer (camera) is moving. We can model this































|  |


in the imaging equations as




⃗⃗



=

2


























=

+  and






⃗⃗







⃗⃗
⃗⃗




























by letting R and   depend on time,   =  ( ),  =  (  ). Assume that the camera is

⃗⃗



⃗⃗













translating with velocity



( ) and that there is no rotation,

( ) = 0 . As we know

from experience, points in the image will seem to move. As we saw in class, image points


⃗⃗












⃗⃗









( ) =

1

















((  

×

)× ).


will appear to move with velocity
































































⃗⃗














The Focus of Expansion (FOE) is the point in the image toward which the camera appears to be
moving, given by the projection of







⃗⃗









|  |
2





into the image.  That is,

















FOE =


⃗⃗




.











⃗⃗











Show that for any world point
















, its image pointwill appear to move with velocity




=   (  



FOE),















proving that image points appear to move toward or away from the FOE. What is k?

More products