INFM346 Обработка на изображения и разпознаване

Анотация:

The course is intended for students in the masters programs in department of Informatics. The aim of the course is to provide an introduction to the basic subjects and principles of two connected disciplines: digital image processing and pattern recognition. The presented topics are fundamental for the field, and the course is designed to be entirely practical, even though it will cover in the required details all the necessary theoretical aspects. Each topic is presented using a system of problems that lead to a working computer program, that is written by the students. Based on this practical approach, the students will also delve into the deep theoretical foundations of digital image processing and pattern recognition.

The topic of digital image processing is addressed in two aspects: improving image quality for visual perception and processing for machine data analysis. The topic of pattern recognition is presented with its application in data extraction from digital images.

The technical environment used during the course is a Linux-based operating system and programming languages C++ and Julia. For the needs of image visualization and manipulation in C++ programs, the framework Qt is used, without shifting the objective of the course. For technical computing, the libraries GNU Scientific Library (GSL) and Fastest Fourier Transform in the West (FFTW) are used.

прочети още
Софтуерни технологии в Интернет

Преподавател(и):

доц. Ласко Ласков  д-р
 Георги Костадинов  

Описание на курса:

Компетенции:

Students successfully finished this course will:

1) know:

• fundamental theoretical background of the digital image processing;

• fundamental theoretical background of the pattern recognition with application to image data.

2) be able to:

• implement the basic image processing algorithms: image filtering, restoration, segmentation, frequency and wavelet analysis;

• implement fundamental methods for image features extraction and representation, statistical image classifiers, neural networks.


Предварителни изисквания:
Students must have the following preliminarly knwolede and skills:

• good knowledge in procedural and object-oriented programming in C++ programming language;

• fundamenal knowledge and understanding in data structures and basic computer algorithms;

• fudamental knwolege and understanding in discrete mathematics, linar algebra, calculus, probability theory.

Форми на провеждане:
Дистанционен

Учебни форми:
Лекция

Език, на който се води курса:
Български

Теми, които се разглеждат в курса:

  1. Introduction to digital images. Input and output operations.
  2. Fundamentals of digital images. Basic relationships between pixels.
  3. Intensity levels transformations. Convolution and filtering in the objective space.
  4. Fourier transform of digital images. Frequency space. Filtering in the frequency space.
  5. Restoration and reconstruction of images.
  6. Color image processing.
  7. Wavelet transforms for digital images.
  8. Image segmentation.
  9. Hough transform.
  10. Feature extraction for the needs of classification.
  11. Statistical methods for classification. Bayes classifier.
  12. Neural networks for image recognition.

Литература по темите:

1. Rafael C. Gonzalez, Richard E. Woods, Digital Image Processing, 4th Edition, Pearson, 2018.

2. Rafael C. Gonzalez, Richard E. Woods, Steven L. Eddins, Digital Image Processing Using Matlab, 3rd Edition, Gatesmark Publishing, 2020.

3. Goodfellow, Ian, et al. Deep Learning. MIT Press, 2016.

4. Milan Sonka, Vaclav Hlavac, Roger Boyle, Image Processing, Analysis and Machine Vision, Cengage Learning, 2015.

5. Duda, Richard O., Hart, Peter E. and Stork, David G., Pattern Classification (2nd Edition), 2 : Wiley-Interscience, 2000.