Computer Vision
Spring 2011

Exam questions

  1. Filters and scale

    • Explain what image filtering is

    • Mention examples of what image filters can be used for

    • Explain the construction of image pyramids (i.e., Gaussian and/or Laplacian)

    • Mention examples of what image pyramids can be used for

  2. Image features

    • Explain how “feature matching” works and mention examples of usage

    • Define what the term “feature detection” means, and explain how the Harris operator works

    • Give an example of at least one “feature descriptor”

    • Give an example of how to obtain scale invariance and/or rotational invariance

  3. Edges and lines

    • Explain how to detect edges using the gradient operator and/or Laplacian operator

    • Explain what is meant by non-maximum suppression and hysteresis (Canny operator)

    • Describe the overall principle of how to extract lines from edge maps

    • Mention examples of usage of edges and lines

  4. Cameras

    • Describe the geometry of the camera model/pinhole camera

    • Define and explain the role of the camera intrinsics and extrinsics

    • Explain the basic principle of camera calibration

    • Give examples of different projection types (e.g., orthographic and perspective)

  5. Projective geometry

    • Describe the basic elements of projective geometry (i.e., projective plane, projective lines, etc.)

    • Explain the difference between a perspective transformation and a projective transformation

    • Explain what is meant by vanishing points and horizon

    • Mention examples of properties/structures that remain invariant – and do not remain invariant – under projective transformation

  6. Stereo and epipolar geometry

    • Explain the epipolar geometry for two calibrated cameras (e.g., epipoles, epipolar lines, essential matrix)

    • Explain what disparity means (in the context of stereo vision)

    • What role does image rectification play in stereo vision?

    • Give at least one example of how to match two stereo images

  7. Motion estimation and warping

    • Explain the basic principle of template matching (i.e., translational alignment)

    • Explain what is meant by hierarchical motion estimation and why this is often useful

    • Describe some basic motion models (e.g., translational, Euclidian, affine, etc.)

    • Mention examples of usage of motion estimation and warping

  8. Optical flow

    • Explain the basic principle of optical flow (i.e., the brightness constancy assumption)

    • Describe at least one algorithm for calculating optical flow (e.g., Lucas and Kanade)

    • Explain what is meant by the “aperture problem”

    • What are the advantages/disadvantages of local motion estimation (e.g., optical flow) vs. global motion estimation (i.e., matching feature points in two images according to some underlying global motion model)

  9. Segmentation

    • Explain the basic concept of snakes (active contour models)

    • Explain how the K-means algorithm works

    • Mention examples of usage of these two techniques

    • Mention examples of other segmentation techniques

  10. Light, color, and 3D reconstruction

    • Explain what is meant by an “environment map” (L)

    • Define the Bidirectional Reflectance Distribution Function

    • Mention examples of how to calculate the light exiting a surface point p under a given lighting condition (diffuse reflection, specular reflection, and phong shading).

    • Explain how photometric stereo works


The oral exam is 25 minutes per student. You start by drawing one of the questions above at random. Then you’ll do your own presentation (15 min.) of the question (no preparation time). In the last 10 minutes, we will ask questions to your presentation or to other parts of the curriculum. After approximately 25 minutes you will leave the room, and the external examiner and I will determine your grade. You will receive your grade immediately afterwards.