Digital Image Processing
Summer Semester 2020, University of Stuttgart
This course is directed to students in their Master’s curriculum and covers digital image processing with applications in computer vision and machine learning. The first part of the course discusses fundamentals in digital image processing including image acquisition and representation techniques, image sampling and quantization, point and morphological image operations, image filtering and correlation, noise reduction and restauration. The second part of the course will present issues and technologies used in modern image processing or computer vision systems. This includes image-based feature extraction and matching, segmentation, motion estimation and classification as well as an overview on state-of-the-art techniques making use of (Deep) Neural Networks. Course units are supplemented by practical application examples from industry. Basic knowledge in linear algebra, probability and system theory, as well as basic programming skills are recommended for this course.
Content overview (preliminary SWS alignment, 28 SWS in total, 14 weeks)
- Introduction and Motivation (2 SWS)
- Image Acquisition and Representation (2 SWS)
- Fourier Transformation and 2D linear systems (2 SWS)
- Image Operations and Image Filtering (3 SWS)
- Feature Extraction and Matching (4 SWS)
- Segmentation (3 SWS)
- Motion Estimation (2 SWS)
- Classification (2 SWS)
- Neural Networks (2 SWS)
- Deep Neural Networks and Applications (3 SWS)
- + Guest Lecture: “Traffic Light Recognition and Parsing using Deep Neural Networks” (1 SWS)
At the end of the course the student will be able to
- Handle techniques of image acquisition, representation and approximation of images in order to extract their meaningful components for a particular application.
- Apply image operations and filters to isolate certain frequency components or to cancel out image noises.
- Extract, register and match structures of interest in an image, such as contours or key features.
- Segment an image into regions of homogeneous characteristics, targeting semantic interpretation of the image content.
- Estimate 3D structures and object motion from video sequences.
- Detect and track objects of interest in single images or video sequences.
- Design and create (Deep) Neural Network Architectures for various tasks, ranging from image segmentation to object detection.
- Identify solutions to complex problems across different application domains, such as quality control, video surveillance, intelligent vehicles and human-machine interfaces.
The course will be based on following textbooks (ordered by prio):
- Computer Vision: Algorithms and Applications, by Richard Szeliski, Springer, 2010. (available online for free, main text book)
- Digital Image Processing, by Bernd Jähne, Springer, 2005. (available online for free)
- Deep Learning, by Ian Goodfellow and Yoshua Bengio and Aaron Courville, MIT Press, 2017 (available online for free)
- Machine Learning: A probabilistic perspective, by Kevin Murphy, MIT Press, 2012
- Probabilistic Robotics, by Sebastian Thrun, Wolfram Burgard and Dieter Fox, MIT Press, 2005
Excercise and exam preparation sessions can be found on ILIAS.
You need a password to unlock the below area including the lecture material (PDFs and Videos). Please contact me to gain access.