Question 1 example 1
Problem 1: Recall (25pts, 5pts each) Directions: Answer the following 5 subquestions. Use only the space provided.
-
What is the fundamental relationship between disparity and depth in a stereo vision system?
-
How many degrees of freedom are there in a 2D Affine transformation?
-
When formulating image segmentation as a graph-based max-flow min-cut problem, what do the two special nodes, the Source ($S$) and the Sink ($T$), represent?
-
True or False: When applying geometric transformations to warp an image, forward transformation mapping is preferred over backward transformation mapping.
-
When evaluating the local uniqueness of an image patch using the eigenvalues ($\lambda_0, \lambda_1$) of the structure tensor, what geometric feature does the window represent if both eigenvalues are large?
Solutions:
-
Solution: Disparity is inversely proportional to depth. (Note: The instructor explicitly stated this was a "fundamental equation to remember" and promised it would be asked on the exam).
-
Solution: 6 degrees of freedom. (Note: The instructor explicitly warned that students would be tested on the degrees of freedom for various transformations).
-
Solution: The Source ($S$) represents the foreground and the Sink ($T$) represents the background.
-
Solution: False. Backward transformation is preferred because forward mapping pushes source pixels into the target grid, which can leave "holes" requiring arbitrary intensity distribution.
-
Solution: A corner.