Computer Vision Quiz

In figure 1, we applied a Gaussian filter to the original image and computed the derivative in x and y directions. Result 1 is the x-derivative of Gaussian, and result 2 is the y-derivative of Gaussian.

• Result 1 is the x-derivative of Gaussian, and result 2 is the y-derivative of Gaussian.

• Result 1 is the y-derivative of Gaussian, and result 2 is the x-derivative of Gaussian.

• None is True

Sort the steps of Marker based Optical Motion capture​:

A.    Joint angle reconstruction​ 

B.    Marker cleanup

C.    Extract markers from video​

D.    Joint position reconstruction​

E.     3D marker position reconstruction​

F.     Track markers over time in video

• A>F>B>E>D>C

• C>F>B>E>A>D

• F>C>B>E>D>A

• C>F>B>E>D>A

Sort the following steps of a Structure from motion​ ​algorithm:

A.    Detect good features​ from many images of the object

B.    Find correspondences between frames​

C.    Estimate Motion and Structure​

D.    Use Bundle adjustment

• A > B > C > D

• D > B > C > A

• A > B > D > C

• A > D > C > B

Given the following flow field: Which of them minimizes the optical flow smoothness energy functional?

• Left optical flow result

• Right optical flow result

What is the human motion capture technology that relies on rotors connected to limb-aligned rods?

• Optical methods

• Magnetic methods

• Elector-mechanical methods

• Marker-based methods

The gradient points in the direction of__________

• lowest increase in intensity.

• most rapid increase in intensity.

• most rapid decrease in intensity.

• lowest decrease in intensity.

What kinds of image regions are hardest to track?

• low texture regions like sky

• Edges of objects

• Corners of objects

• Rotating spheres

Why does the first stage of any edge detection algorithm involve convolution with a blurring mask?

• To increase the image sharpness

• To reduce the effects of noise

• To reduce the number of pixels

• To detect edges

When using Harris detector, an image region is considered to be a corner if:

• both λ1 and λ2 are small.

• both λ1 and λ2 are large.

• both λ1 greater than λ2.

• both λ1 equal than λ2.

Gradient magnitude images are more useful in:

• point detection

• line detection

• area detection

• edge detection