Communication and Media Engineering

Bewerbung
Fakultät EMI

Modulhandbuch

←  Zurück 

Computer Vision

Prerequisite

Linear Algebra
Teaching methods Lecture/Lab
Learning target / Competences

Target skills:

The student will gain an overview on established and modern image processing techniques. The course provides tools, methods, models and techniques for the following topics: image formation, optics, imagers, color, image segmentation, image analysis, image features, image alignment, estimation in computer vision, programming and deep learning.

 

Competences:

The student will understand basic problems in image processing and machine vision, e.g. image segmentation, feature detection, image matching or estimation problems in alignment.

He/she will know methods, algorithms and common techniques to solve the above mentioned problems.

The student will be able to computationally apply the methods on given low-level and higher-level image processing tasks in real world computer vision problems.

 
Duration 1
Hours per week 4.0
Overview
Classes 60 h
Individual / Group work: 60 h
Workload 120 h
ECTS 4.0
Requirements for awarding credit points

Computer Vision with Lab

Written exam K60+Lab

Das unbenotete Labor ist Voraussetzung für die Zulassung zur Klausur K60.
Credits and grades

4 CP, grades 1 ... 5
Responsible person

Prof. Dr.-Ing. Stefan Hensel
Recommended semester 2
Frequency Every 2nd sem.
Usability

Master's degree program CME, EIM and MMR
Lectures

Maschinelles Sehen mit Labor

Type Lecture/lab
Nr. EMI2247
Hours per week 4.0
Content

Lecture contents:

Feature-based methods:

  • Feature detectors and feature descriptors
  • SIFT detector and descriptor

Image Transformations:

  • Affine and Projective Transformations
  • Robust transformation estimation (RANSAC)

Image Motion and Tracking

  • Visual odometry and optical flow (local and global methods)

Machine learning in image processing

  • Clustering/Segmentation: k-means, SLIC Superpixel, spectral methods
  • Classification: Support Vector Machines

Deep learning in machine vision

  • Fundamentals of deep neural networks in image processing (convolutional neural networks, CNNs)
  • Training and training data collection
  • Object classification with neural networks
  • Object detection and segmentation with neural networks

 Laboratory contents:

  • image mosaicing: image transformations and scale-invariant feature detectors
  • Visual Odometry: Non-contact speed determination in video sequences
  • Machine learning methods for segmentation: K-Means in image compression
  • Deep Learning: Object classification and detection
  • Deep Learning: Keras, Tensorflow and python-based open source usage

 Literature:

  • Szeliski, R., Computer Vision: Algorithms and Applications; Springer, 2011, online pdf version: http://szeliski.org/Book/
  • Burger, Burge, Digital Image Processing - An algorithmic introduction, 3rd ed. Springer, 2015
  • Gonzalez, Digital Image Processing, 4th ed., Pearson, 2017
  • Goodfellow, Bengio, Courville, Deep Learning, MIT Press 2016, onlineversion: http://www.deeplearningbook.org/

 
Literature
  • Szeliski, R., Computer Vision: Algorithms and Applications; Springer, 2011, online pdf version: http://szeliski.org/Book/
  • Burger, Burge, Digital Image Processing - An algorithmic introduction, 3rd ed. Springer, 2015
  • Gonzalez, Digital Image Processing, 4th ed., Pearson, 2017
  • Goodfellow, Bengio, Courville, Deep Learning, MIT Press 2016, onlineversion: http://www.deeplearningbook.org/
←  Zurück