TU Wien:Computer Vision VU (Sablatnig)/Prüfung 2021-09-30

3D Reconstruction

• What is Structure from Motion?
• Given a fundamental matrix relating the images 1 and 2 and a point x in image 1, what can be said about the corresponding point in image 2?

Deep Learning

• What is the difference between the “classical” machine learning pipeline for image recognition and new deep learning methods?
• Explain the concept and usage of auto-encoders.

Gabor Wavelets

• What are Gabor Wavelets?
• Describe the difference between the real and imaginary component of Gabor Wavelets.

Harris Corner Detection

• How is the “cornerness” of an image pixel computed by the Harris corner detection algorithm?
• How are the final corner points detected from the cornerness?

Homography and RANSAC

• Given a set of putative matches between two images of a panorama series: describe how the RANSAC algorithm could be used to estimate the homography between the images.
• Imagine 1000 matches are given between the two images, but only 300 of them are correct. Will RANSAC likely be able to estimate the correct transformation? Justify your answer.

Image Classification and Machine Learning

Write the number of the statement and if it is ‘incorrect’ or ‘correct’ (a wrong answer counts -1 point).

• The bias and variance of a classifier are independent of each other
• In k-means clustering the value k is automatically determined by the algorithm
• On average, images of man-made environments have more vertical and horizontal edges than images of natural environments
• In the bag of words model, regular sampling of interest points can be used instead of interest point detection
• The dimensionality of a bag-of-words histogram is 128 when SIFT is used for local feature extraction
• Supervised learning is also called “clustering”

Image Pyramids

• What are Gaussian pyramids and how are they generated?
• How can a Laplacian pyramid be computed from a Gaussian pyramid?

Bag of Words

• What is clustering used for in the bag-of-words model?
• Explain the Spatial Pyramid Matching (SPM) method in bag-of-words image classification.

SIFT

• How is scale invariance achieved in SIFT keypoint detection?
• How is rotation invariance achieved in SIFT keypoint detection?

Thin Lens

• What is described by the thin lens equation?
• Describe all variables of the equation.