# Krystian Mikolajczyk

### Image Processing and Vision

MSc module

Aims/Learning Outcomes

• The aim of this module is to offer an in depth course on the basic principles of Image Processing and Vision which
may form the foundation for a variety of diverse disciplines like Remote Sensing, Robot Vision, Image and Vision
Understanding, Medical Imaging, Digital Broadcast etc. Indexing and retrieval of images and videos, image databases,
image search engines, video exploration and annotation
.

• By the end of this module the student will:

• have a systematic understanding of image processing and computer vision issues,

• be able to formulate problems in a Mathematical way and solve them to achieve optimality in performance,

• be able to analyse complex problems in digital image processing, understand the concepts behind them and
come up with possible solutions.

Content

• Introduction - Definition of an image, digitisation, criteria for sampling and quantisation.
Image Transformations - Matrix and vector representation of images. Orthonormal bases. Linear operators.
2D unitary transforms. Singular Value Decomposition of matrices. 2D Finite Fourier, Walsh, Hadamard and
Haar transforms. The image as a random field. Karhunen-Loeve transform and principal component analysis.
Finding correspondences between images.

• Image Enhancement and Filtering - histogram modification, smoothing, sharpening.
Scale Space - basic scale-space theory, Gaussian kernel and its derivatives, scale-space pyramids, interpolation.

• Feature Detection - basic image structures, features, feature detection algorithms, image descriptors, similarity measures.
Texture - texture models, descriptors, classification and segmentation.

• Geometry - geometric image transformations, homogenous coordinates, matching algorithms, RANSAC.
Segmentation - thresholding, watershed algorithm, texture segmentation, clustering algorithms, segmentation algorithms.

• Human Vision System - physiology of the human vision system. psychophysical experiments, results and implications,
visual perception, colour.

• Motion Analysis - motion models, tracking methods, motion based segmentation.
Image registration -

• 2D Shape - detection, description, representation, Hough transform.
3D Shape - multiview representation, mesh, multiscale, geodesic lines.
Geometry of Vision - camera model, camera calibration, stereo vision.

• Problem Class

Method of teaching

• Lectures: 10weeks, 3 hours per week

• Labs: A lab exercise to consolidate the lecture material, 1 day

Assessment

• Examination - 2 hour unseen paper. Answer 3 questions out of 4. - 85%

• Lab Report - Answer all questions in the lab specification. - 15%

• Part-time Students: Examination counts for 100%.

Selected texts

• Gonzales RC., Woods P., Digital Image Processing, 2002. 0-201-600781

• Gonzales RC., Woods P., Eddings, Digital Image Processing using Matlab, 2004

• Petrou,M., and Bosdogianni, P., Image Processing: the fundamentals, 2000.

• Nixon, M., and Aguado, A., Feature Extraction and Image Processing, 2008.