Course detail

Advanced Methods in Image Processing

FEKT-MPA-AB2Acad. year: 2024/2025

The subject is designed as an extension of the previous subject Image processing and analysis, which is taught in the 3rd semester of the master's study program. The form of teaching is project-based, where students solve assigned tasks from various areas of image data processing within selected teams. Specifically, these areas are: image noise suppression, image restoration, landmark detection and feature extraction, stereoscopy, camera calibration methods, disparity map estimation, 3D object reconstruction, advanced methods for image matching, object tracking, and optical-based motion detection flow, image segmentation.

 

Language of instruction

English

Number of ECTS credits

5

Mode of study

Not applicable.

Guarantor

Ing. Jiří Chmelík, Ph.D.

Department

Department of Biomedical Engineering (UBMI)

Offered to foreign students

The home faculty only

Entry knowledge

The following knowledge is required to enrol in this course:

1) Processing and analysis of signals (theory of analogue and digital signals, filtering, Fourier and wavelet transformation, spectral analysis).

2) Processing and analysis of images and other multidimensional signals (theory of nD signals, image restoration methods, image segmentation methods, texture analysis, image data reconstruction methods).

3) Basic knowledge of machine learning methods and statistical analysis (linear classifiers, clustering methods, neural networks, SVM, PCA, probability theory).

4) Mathematics at the technical college level (derivatives, integrals, solving integrodifferential equations, optimization tasks).

5) Advanced programming experience (MATLAB or Python)

The main prerequisite is the successful completion of the previous course MPA-ABO (Analysis of Biomedical Images). The composition of the MPA-AB2 course is closely related to the material discussed in this course (MPA-ABO). 

Rules for evaluation and completion of the course

The course runs in blocks - 7 weeks / 7 hours + 3 hours. Lectures are conducted in the form of mandatory seminars, which immediately follow on with mandatory computer exercises (solving joint projects in groups in the form of hackathons). 

The conditions for successful completion of the course are determined by the annually updated Announcement of the subject guarantee.

- full participation in lectures and exercises

- preparation and delivery of a presentation on a given topic

- success in the final exam (the condition for passing the course is to obtain at least 50 points in total) 

Aims

The course aims to familiarize students in the last semester of the master's study program with selected advanced methods in the field of image processing and computer vision, which are applicable to a wide range of applications. The goal is to acquire the appropriate theoretical basis of the discussed methods and, within team projects, to be able to practically apply the acquired knowledge in order to solve the selected task.

 

Study aids

Not applicable.

Prerequisites and corequisites

Not applicable.

Basic literature

JAN, Jiri, 2019. Medical Image Processing, Reconstruction and Analysis: Concepts and Methods. Second Edition. Boca Raton: CRC Press. ISBN 9781138310285. (EN)
Kundur, D., Hatzinakos, D.: Blind image deconvolution, IEEE Signal processing magazine, 1996, pp. 43-64 (EN)
Rudin, L. I. et al.: Nonlinear total variation based noise removal algorithms, Physica D vol. 60, 1992, pp. 259-268 (EN)

Recommended reading

Not applicable.

Classification of course in study plans

  • Programme MPA-BTB Master's 2 year of study, summer semester, compulsory-optional
  • Programme MPC-BTB Master's 2 year of study, summer semester, compulsory-optional
  • Programme MPC-BIO Master's 2 year of study, summer semester, elective

Type of course unit

 

Exercise in computer lab

52 hod., compulsory

Teacher / Lecturer

Ing. Jiří Chmelík, Ph.D.

Syllabus

1. Advanced methods for noise suppression (basic methods and advanced approaches - anisotropic diffusion, total variation, deep learning).
2. Selected image restoration methods (distortion models, blind deconvolution, Tikhonov regularization, deep learning).
3. Detection of keypoints, local feature extraction, point matching (SIFT, SURF, and others).
4. Stereoscopy, multiscopy, camera calibration methods, disparity map estimation, reconstruction of 3D objects.
5. Advanced image registration methods (flexible approaches, mark correspondence, ICP method, Elastix program).
6. Object tracking and motion detection methods based on optical flow.
7. Advanced image segmentation methods (graph-based methods and Markov random fields).