DIZO

Digital image processing : FNSPE-CTU in Prague

COURSE DESCRIPTION

Foundational Training in Image processing and Pattern Recognition

This course emphasizes the conversion of images into digital form, initial processing techniques like noise reduction, improving contrast, and removing blur, as well as techniques for identifying edges, dividing images into segments, and applying geometric changes. Theoretical concepts will be illustrated through experimental examples and real-world uses. Practical sessions will be held in computer laboratories, with coding exercises in Python.

VICE CHANCELLOR

COURSE SCHEDULE

15.2. - 10:00 - 13:30 CANCELED

22.2. - 10:00 - 13:30 labs ON-LINE

29.2. - 10:00 - 13:30 lecture

07.3. - 10:00 - 13:30 lecture

14.3. - 10:00 - 13:30 lecture

21.3. - 10:00 - 13:30 labs ON-LINE

28.3. - 10:00 - 13:30 lecture

04.4. - 10:00 - 13:30 labs ON-LINE

11.4. - 10:00 - 13:30 lecture

18.4. - 10:00 - 13:30 lecture

25.4. - 10:00 - 13:30 labs ON-LINE

02.5. - 10:00 - 13:30 labs ON-LINE

09.5. - 10:00 - 13:30 labs ON-LINE

16.5. - CANCELED [Wednesday timetable]

Classes take place only in the summer semester. Time and room according to the FNSPE schedule.

COURSE OUTLINE

  • Convolution and Fourier transform in the continuous and discrete domain

  • Digitization of signals and images - Sampling theorem, Nyquist inequalities, reconstruction of continuous signal under sampling, interpolation, quantization, quantization noise

  • Histogram and its transformation (equalization, contrast enhancement), color in the image

  • Models of noise in the image and methods for its suppression (convolution filters, median, bilateral filter, NLM, others)

  • Edge detection in the image (derivative methods, frequency domain, Hough transform)

  • Basic types of image blurring, their modelling and estimation, inverse and Wiener filters

  • Geometric registration (matching) of images -basic principles and methods (image and phase correlation, transform models, resampling)

  • "Handcrafted" features for 2D objects I - visual features, Fourier descriptors, local features, SIFT

  • "Handcrafted" features for 2D objects II - Moments, moment invariants due to rotation and scaling, normalization

COURSE OUTLINE

  • Lab 1:

    • Introduction to Python

    • Python fundamentals for Image Processing

    • Colab, GitHub

    • Basic Image Manipulation

    • Fourier Transform

  • Lab 2:

    • Noise Reduction

    • Edge detection

    • Histogram

  • Lab 3:

    • Image Registration

    • Morphological

    • Transformations

  • Lab 4:

    • Hough and Radon Transform

  • Lab 5:

    • Segmentation by Thresholding

    • Object Recognition

  • Lab 6:

    • Digital image processing in practice

COURSE HANDOUTS

The files contain most of the slides used. This is not the complete lecture content; the knowledge recorded in the files is insufficient to pass the exam!

RECOMMENDED LITERATURE

[1] Pratt W. K.: Digital Image Processing (3rd ed.), John Wiley, New York, 2001

[2] Gonzales R. C., Woods R. E., Digital Image Processing (2nd ed.), Prentice Hall, 2002

[3] Zitová B., Flusser J., Image registration methods: a survey. Image and Vision Computing, 21 (2003), 11, pp. 977-1000

EXTENSION MATERIAL

LABS

We are going to take a step into the world of Image Processing using python. In order to do that, you should know the basics of Python. Please go through the following section before attending labs. Necessary information for completing all labs will be provided here. If you have any questions please contact novozamsky@utia.cas.cz

The exercises will be in Czech, but all materials and texts will be in English.

REQUIREMENTS

Zoom account - the webcam will be on !

GitHub account

Gmail account for Colab is recommended

Browser

PYTHON CRASH COURSE

LABS START HERE!

At the beginning of every lab session, join Zoom Meeting, where you receive a link for a GitHub classroom assignment. It contains all codes for online/offline studying.

  1. Zoom link: https://cesnet.zoom.us/j/96991538006

  2. Get a link from your teacher, e.g. https://classroom.github.com/a/YspbK****

  3. Sign in to GitHub

  4. If necessary, click on the "Authorize GitHub Classroom" button

  5. Find your name in the list (only for the first assignment)

  6. Click on "Accept the assignment"

  7. Go to colab.research.google.com

  8. Click on "Authorize Google Colab"

  9. Click on "GitHub" (or File > Open Notebook > GitHub)

  10. Type your GitHub user name (check "Include private repos")

  11. Find the desired repository and the notebook file

  12. Loading (In case of "Error", press "Retry")

  13. If necessary, click on the "Authorize with GitHub" button

  14. Edit the notebook, finish excercises

  15. Save in GitHub (File > Save a copy in GitHub)

MATERIALS FOR LABS

Lab 1 - DIZO001 - Digital Image Processing - 16. 02. 2024 - 10:00 a.m.

Introduction to Python, Python fundamentals for Image Processing, Colab, GitHub, Basic Image Manipulation, Fourier Transform
GitHub Classroom link

Lab 2 - DIZO002 - Digital Image Processing - 21. 03. 2024 - 10:00 a.m.

Noise Reduction, Edge detection, Histogram
GitHub Classroom link

Lab 3 - DIZO003 - Digital Image Processing - 04. 04. 2024 - 10:00 a.m.

Hough and Radon Transform
GitHub Classroom link

Lab 4 - DIZO004 - Digital Image Processing - 25. 04. 2024 - 10:00 a.m.

Image Registration, Morphological Transformations
GitHub Classroom link

Lab 6 - DIZO006 - Digital Image Processing - 02. 05. 2024 - 10:00 a.m.

Digital image processing in practice

GitHub Classroom link

Lab 5 - DIZO005 - Digital Image Processing - 09. 05. 2024 - 10:00 a.m.

Segmentation by Thresholding and Object Recognition
GitHub Classroom link

EXAMS

Search