Nptel Online Certification Course

Indian Institute of Technology Kharagpur

Computer Vision
Assignment - Week 4
Number of questions: 10 Total marks: 10x2=20

QUESTION 1: Type: MCQ

Given the right epipole 𝑒′ (2, 3) and a point 𝑥 ′ (5, 2) in right image plane of a stereo imaging
set up, compute the epipolar line 𝑙 ′ containing the point 𝑥 ′, in right image plane in
homogeneous coordinate system.

a) (0.0909, −0.2727, 1)

b) (0.0909, 0.2727, 1)

c) (−0.0909, 0.2727, −1)

d) (−0.0909, −0.2727, 1)

Correct Answer: : d)
Detailed Solution: Epipolar line 𝑙 ′ = 𝑒′ × 𝑥 ′ in homogeneous coordinate system.

1
QUESTION 2: Type: MCQ
Assume that a stereo imaging setup has two image planes, left and right. Which of the
following could be a possible right null vector (or left epipole for the corresponding stereo
 20 12 0 

imaging setup) of the given fundamental matrix 𝐹 =  8 −7 59  .
−4 33 −177
 
a) (−3, −5, 1)

b) (3, 5, 1)

c) (−3, 5, 1)

d) (3, −5, 1)

Correct Answer: : c)
Detailed Solution: Check for 𝐹𝑒 = 0, where 𝑒 is the left epipole or right null vector of 𝐹.

2
FOR QUESTIONS 3, 4 AND 5:
1 1 2
 
Consider a stereo imaging set up camera with left calibration matrix 𝐾 = 0 2 0 and a
 
0 0 1
 
1 5 3 1 0 0
  
right calibration matrix 𝐾 ′ = 0 1 2 . The right camera has 𝑅 = 0 0 1
0 0 2 0 1 0
  
Based on the given parameters, solve the following question 3, 4 and 5:
QUESTION 3: Type: Comprehensive
Find the homography matrix 𝐻.
1 1 3
 
a) 3 1 1
2 0 0
 
1 1 3
 
b) 0 1 1
0 1 0
 
1 2 3
 
c) 0 1 2
5 0 3
 
1 5 3
 
d) 2 1 2
3 3 2
 
Correct Answer: b)
Detailed Solution:
Homography is given by 𝐻 = 𝐾 ′ 𝑅𝐾 −1 .

3
QUESTION 4: Type: Comprehensive
Given translation vector 𝑡 as (1, 0, 1), compute the right epipole.

a) (4, 2, 2)

b) (2, 2, 2)

c) (4, 3, 2)

d) (4, 2, 1)

Correct Answer: a)
Detailed Solution:
Right epipole is given by 𝑒′ = 𝐾 ′𝑡.

4
QUESTION 5: Type: Comprehensive
Compute the fundamental matrix 𝐹.
−1 0 −2
 
a)  1 0 6 
 0 2 −2
 
−2 2 −2
 
b)  2 −2 6 
−2 2 −2
 
 0 0 −1
 
c)  1 −1 6 
−1 1 −1
 
 0 0 −2
 
d)  2 −2 6 
−2 2 −2
 
Correct Answer: d)
Detailed Solution:
Fundamental matrix is given by 𝐹 = [𝑒′] 𝑋 𝐻.

5
FOR QUESTIONS 6 AND 7:
−1 3 2
 
Given fundamental matrix 𝐹 =  1 1 0
 1 1 0
 
Based on the given data solve the following questions 5 and 6:
QUESTION 6: Type: Comprehensive
Given two points (5, 8) and (7, −5) in the left image. Compute the epipolar lines.

a) (−21, 13, 13), (−20, 2, 2)

b) (21, 13, 13), (−20, 2, 2)

c) (21, 13, 13), (20, 2, 2)

d) (−21, 13, 13), (20, 2, 2)

Correct Answer: : b)
Detailed Solution: Epipolar line is given by 𝑙 ′ = 𝐹𝑥

6
QUESTION 7: Type: Comprehensive
Find the right epipole.

a) (0, 302, 302)

b) (0, −302, −302)

c) (0, 302, −302)

d) (0, −302, 302)

Correct Answer: : d)
Detailed Solution: Epipolar line is given by 𝑒′ = 𝑙 1′ × 𝑙2′

7
QUESTION 8: Type: MSQ
1 0 0 0
 
Consider a stereo imaging set up with two cameras 𝑃 = 0 1 0 0 (left camera) and 𝑃′ =
0 0 1 0
 
1 2 5 1
 
2 0 0 2 (right camera), which of the following options are correct?
 
3 0 2 1
 
4 0 0 

a) 𝐹 = 2 2 −3 

0 −4 −10
 
b) The right epipole is (1, 2, 1).
4 0 4 

c) 𝐹 = −2 2
 3 
 0 −4 −10
 
d) The right epipole is (0, 0, 1).

Correct Answer: : b, c
Detailed Solution: Right epipole will be 𝑒′ = 𝑃′𝐶. Fundamental matrix 𝐹 = [𝑚] 𝑋 𝑀, where
𝑚 is the last column of 𝑃′ and 𝑀 is the left 3x3 matrix of 𝑃′.

8
QUESTION 9: Type: MCQ
1 0 0
 
Given the right camera as 𝑅 = 0 0 1 and translation vector 𝑡 as (1, 1, 1), compute the
0 1 0
 
Essential matrix 𝐸.
0 1 1
 
a) 1 1 0
1 0 2
 
−1 −1 0
 
b) −1 0 2
 0 −1 1
 
 0 1 −1
 
c)  1 −1 0 
−1 0 1 
 
−2 −1 0
 
d) −1 1 1
 0 −1 1
 
Correct Answer: c)
Detailed Solution:
Essential matrix is given by 𝐸 = [𝑡 ′] 𝑋 𝑅.

9
QUESTION 10: Type:Numeric
2 0 0 0
 
Consider a stereo imaging set up with two cameras 𝑃 = 0 2 0 0 (left camera) and 𝑃′ =
0 0 1 0
 
2 0 0 5
 
0 2 0 0 (right camera). If the image coordinates of a 3-D point are (10, 3) and (15, 3)
 
0 0 1 0
 
in left and right cameras, compute its depth (z-coordinate) in the 3D. (Answer only the
magnitude of the depth)
Correct Answer: 1
Detailed Solution:
2× 25
The depth is given by 𝑍 = 𝑥𝐾𝑡′ −𝑥 = 15−10 .

10