Computer Vision 1 (Winter Term 2021/2022)

Overview

• Online course (2/2/0) consisting of:
  • Live video lectures with Q&A on Wednesdays, 11:10 - 12:40 → Join here
  • Live online exercises on Mondays, 14:50 - 16:20 → Join here
  • Assignments, self-study and moderated discussion in a forum
• Lecturer: Bjoern Andres
• Teaching Assistant: Holger Heidrich
• Enrolment (OPAL). Additional rules for enrolment may apply, depending on the study programme.
• Creditable toward the modules CMS-CE-EL1, CMS-CE-EL2, CMS-CLS-ELG, CMS-VC-ELG, CMS-VC-ELV1, CMS-VC-ELV2, INF-04-FG-IS, INF-B-510, INF-B-520, INF-B-530, INF-B-540, INF-BAS2, INF-BAS7, INF-E-3, INF-LE-MA, INF-VERT2, INF-VERT7, INF-VMI-8, INF-VMI-8A, MATH-MA-INFGDV, MCL-AI
• Examinations (announced on 2021-12-22): This term, we will offer remote oral examinations only for this course. In order to take this examination, eligible students need to complete all of the following steps:
  • Students enrolled in the study program Computational Modelling and Simulation (CMS) need to:
    • register for the examination via SELMA before January 16th
    • schedule (choose) an appointment for the examination through OPAL before January 25th
    • send to us via this email template before January 25th:
      • the filled-in and signed declaration of consent to an alternative form of performance of a non-objective examination
      • the filled-in spreadsheet file
  • Students enrolled in other degree programs offered by the faculty of computer science need to:
    • register for the examination with the examination office before January 16th, as described here
    • schedule an appointment for the examination before January 25th, by sending the confirmation of the registration to the secretary of the main examiner. If Bjoern Andres is to act as the main examiner, send to us via this email template before January 25th:
      • the confirmation of the registration
      • the filled-in and signed declaration of consent to an alternative form of performance of a non-objective examination
      • the filled-in spreadsheet file

    These deadlines are strict. Students who are not registered or have not scheduled an appointment will not be able to take the examination this term.

Contents

• Lectures
  • Formation of digital images
    • Geometric optics
    • Aberration
    • Sensor types and technologies
    • Noise
    • Digital images
    • Color spaces
  • Operators on digital images
    • Point operators
      • Gamma correction
      • Histogram equilibration
      • Tone mapping
    • Linear operators
      • Convolution
      • Averaging filters
      • Derivative filters
      • Convolutional neural networks
    • Non-linear operators
      • Median filter
      • Bilateral filter
      • Morphological filters
      • Distance transform
    • Classic computer vision
      • Canny edge detector
      • Harris corner detector
      • Scale-invariant feature transform (SIFT)
  • Fourier analysis of digital images
    • Discrete Fourier transform
    • Convolution
    • Correlation
    • Sampling theorem
    • Discrete cosine transform
    • Uncerteinty principle
  • Geometric analysis of digital images
    • Projective geometry
    • Panorama stitching
    • Epipolar geometry
    • Stereo vision
  • Classification of digital images (slides)
    • Classification of image patches
    • Smoothness prior
  • Segmentation of digital images (slides)
    • Multicuts
    • Semantic segmentation
  • Recognizing objects in digital images (slides)
    • Object recognition as joint graph decomposition and node labeling