17PMPZAO, 17PMBZAO, 17AMBZAO Image Processing and Analysis

For pages of the subject in Czech, click here.

Subject finishing	Credit, examination
Semester	Winter
Subject extend	2 hours lecture + 2 hours labs
Credits	5
Type of study	Full time
Responsible	Prof. Ing. Václav Hlaváč, CSc., FEL, Ing. Zoltán Szabó, PhD.
Department	Department of Biomedical Technology

Anotace

The subject aims to teach students image processing and analysis methods. Image processing deals with the situation, in which no semantic information about the image content is available. Image analysis can explore semantics. The image can be segmented into objects of interest. The rest of the image is treated as uninteresting background. Having objects, enables describing them and recognizing them using statistical decision making methods. We will build on top of students knowledge in calculus, linear algebra and signal theory.

Lectures - contents and schedule

The lectures are read on Thursdays from 10:00 to 11:40 at FBMI ČVUT in Kladno, room KL:B-701

Lecturers: prof. Ing. Václav Hlaváč, CSc., vaclav.hlavac@cvut.cz, Czech Institute of Informatics, Robotics and Cybernetics, CTU in Prague
Ing. Zoltán Szabó, Ph.D, szabo@fbmi.cvut.cz, Faculty of Biomedical Engineering, ČVUT in Kladno, Department of Biomedical Informatics

Week of the semester	Date	Lecturer	Lecture content
1	5.10.2017	V. Hlaváč	Image processing vs. computer vision. Role of interpretation. Objects in the image. Digital image. Distance transform.
2	12.10.2017	Z. Szabó	Brightness transformation, geometric transformation, image interpolation
3	19.10.2017	V. Hlaváč	Image acquisition from both geometric and radiometric view..
4	26.10.2017	V. Hlaváč	Fourier transform in 1D and 2D. Image filtration in the frequency domain.
5	2.11.2017	V. Hlaváč	Image processing in the spatial domain. Convolution, correlation. Noise filtration.
6	9.11.2017	V. Hlaváč	Edge detection. Scale space. Salient point detection in the image.
7	16.11.2017	V. Hlaváč	Principal compotent analysis method. Wavelet transformation.
8	23.11.2017	V. Hlaváč	Segmenting objects in images. Thresholding. K-means. Mean shift.
9	30.11.2017	V. Hlaváč	Optimization tasks and methods. An overview.
10	7.12.2017	V. Hlaváč	Segmenting objects in images formulated as optimization tasks. Object description.
11	14.12.2017	V. Hlaváč	Statistical pattern recognition. Role of learning. Use for images.
12	21.12.2017	V. Hlaváč	Mathematical morphology. Both binary and grayscale methods.
13	4.01.2018	V. Hlaváč	Image compression. Color images.
14	11.01.2018	V. Hlaváč	Rigid and non-rigid registration of images and objects in them.

Web pages of labs

Instructors of labs are
Ing. Jan Hejda, Department of Biomedcal Technology FBMI, jan.hejda@fbmi.cvut.cz for the subject 17PMBZAO
Doc. Ing. Zoltán Szabó, Ph.D., szabo@fbmi.cvut.cz Department of Biomedical Informatics for the subjects 17PMPZAO, 17AMBZAO.
Instructors maintain web pages for the labs.

Examination and its evaluation

Only students who obtained the credit for their lab activity are eligible for the examination.
The examination consists of two parts, written and oral exams. The written part checks the global orientation of the student in the subject matter. Students typically answers six questions, which are randomly selected from the fixed list of questions. The written exam lasts 40 minutes. The written part of the exam yields 30 points at maximum.
The oral part of the exam follows the written part after the written part is correctd by the teacher. The oral part is a discussion of a student and the teacher about a scientific paper of student's choice. The paper has to be from a respected scientific journal, which cannot be older than five years. The paper has to have relation to the subject and be written in English. The priviledge to choose the paper gives the student the oportunity to bring the discussion to the area he has a deeper knowledge. Students comes to the exam with a printed version of the paper with her/his hand written notes made while reading the paper.
The list of journals from which the paper can be selected: IEEE Transactions on Pattern Analysis and Machine Intelligence, IEEE Transaction on Medical Imaging, International Journal on Computer Vision, Medical Image Analysis. The Czech Government pays to its universities the electronic access to papers. Use https://dialog.cvut.cz/
The examination mark is given by the sum of points. Labs (max. 40 points), written part (max. 30 points) and oral exam (max. 30 bodů).
The maximal number of points equals to 100. Examination results: A 100-90 points, B 89-80 points, C 79-70 points, D 69-60 points, E 59-50 points, F < 50 points.

Recommended literature