Image processing 2400-ZEWW988

The course will sequentially cover the following topics:

1. Image preprocessing:
conversion to greyscale, normalisation, contrast enhancement (e.g. histogram equalisation)
noise reduction using Gaussian and median filtering,
resizing, sharpening, blurring,
data augmentation techniques such as rotation, flipping, cropping, and random colour alterations (colour jittering).

2. Geometric transformations:
basic transformations including translation, scaling, rotation, and affine transformations,
perspective transformations and homography,
interpolation methods: nearest-neighbour, bilinear, and bicubic.

3. Morphological operations:
processing of binary images,
erosion, dilation, opening, and closing,
hit-or-miss, top-hat, and black-hat operations.

4. Edge detection and filtering:
filtering fundamentals: box filter, Gaussian, and median filters,
edge detection algorithms: Sobel, Prewitt, Laplacian,
Canny edge detector: gradient calculation, non-maximum suppression, and hysteresis thresholding,
low-pass vs. high-pass filtering.

5. Feature extraction:
classical descriptors: corner detectors – Harris, Shi-Tomasi; keypoint detectors – SIFT, SURF, ORB,
descriptors and matching: BRIEF, FREAK, feature matching using FLANN.

6. Image segmentation:
semantic and instance segmentation,
thresholding (Otsu’s method), the Watershed algorithm,
region-growing methods,
U-Net architecture and its variants.

7. Frequency-domain transformations:
Fourier transform (DFT, FFT): amplitude and phase analysis
low-, high-, and band-pass filtering,
Discrete Cosine Transform (DCT),
Wavelet Transform: multi-resolution analysis and denoising.