Image Analysis and Object Recognition

The course gives an introduction to the basic concepts of pattern recognition and image analysis. It covers topics as image enhancement, local and morphological operators, edge detection, image representation in frequency domain, Fourier transform, Hough transform, segmentation, thinning, object categorization and machine learning for visual object recognition.

The course is managed via the learning platform Moodle. All documents and further information can be found in the Moodle course Image Analysis and Object Recognition SoSe2024.

Please notice: The materials for our lectures and exercises are only available through the network of the Bauhaus-Universität Weimar.

Lecture

Introduction

  1. Organization, motivation, definition, application examples (slides, print)

Image processing

  1. Image representation, basic enhancement, binary images, thresholding, morphological operators
  2. Boundaries, filling, hit-or-miss, thinning, thickening, connected components, region properties
  3. Invariant descriptors, (non-)linear filtering, noise, local operators, Gaussian, image derivatives, Sobel
  4. Laplacian, ...

Feature extraction

  1. Gaussian derivatives, GoG, LoG, DoG, edge detection, corner detection, interest point operator
  2. Discrete 2D Fourier transform, FFT, spectrum, convolution theorem, filtering in frequency domain, ideal high- and low-pass filter, Butterworth

Shape detection

  1. Hough transform, fitting of straight lines, line segments, circles with (un)known radius, generalized Hough transform, Fourier descriptors, normalization

Object recognition

  1. Sliding window, template matching, image pyramids, Viola-Jones face detection, Haar features, integral images, boosting, classifier cascade, local image features
  2. Scale invariant extraction, scale-space, unique feature description and matching, SIFT
  3. Part-based voting, vocabulary tree, bag of visual words, ...

Image regions

  1. Segmentation, clustering, k-means
  2. Mean shift, region growing, watershed, graph-cut segmentation, affinity matrix, normalized-cut, soft segmentation, grab-cut, ...

Machine learning

  1. Pattern recognition systems, supervision, generalization
  2. Deep learning for visual recognition, artificial neural networks, supervised convolutional nets
  3. Classifiers, k-nearest neighbor, boosted decision tree, Bayesian decision theory, expectation maximization, support vector machine, summary

Exercise

The first exercise class will take place on 18th of April.

All documents and further information can be found in the <link https: moodle.uni-weimar.de course>Moodle course Image Analysis and Object Recognition SoSe2024.

MATLAB primer (Mathworks, University of Florida)

Exam

Written exam

  • Date: to be announced
  • Auxiliary resources: none

Preparation material