3 ECTS credits
90 h study time
Offer 1 with catalog number 4010976ENR for all students in the 2nd semester at a (E) Master - advanced level.
The fundamental grounds in digital image processing are set by linear algebra, digital signal processing and statistics. Due to insufficient knowledge of the imaged scene, it remains an art on top of a scientific discipline to highlight the relevant information or to extract it from images. This extracted information varies depending on the goals that are pursued or the application field (context) that is considered. This course is focused on processing of measured and discretized image data, without taking into account a priori contextual models of the scene.
1) Global Image Transforms: General Model; Physical Meaning of the expansion in basis images; How to calculate the coefficients?; Separable Transforms; Direct, Forward Transforms, Orthonormal Case; Discrete Karhunen Loeve Transform - KLT; Proofs and Construction of KLT basis images; Application of KLT for Image Compression; Discrete Fourier Transform (1D, 2D definitions, existence, terminology, properties, display of Fourier coefficient images); Discrete Cosine Transform – DCT (1D, 2D definitions), application of DCT in image compression.
2) Wavelet Transform: Drawbacks of the Fourier Analysis; Time-Frequency Representations, uncertainty principle; Continuous Short-Time Fourier Transform; Continuous Wavelet Transform; Frames; The Multiresolution Representation; Integer Wavelet transform and the lifting scheme; Application Examples (Embedded Zerotree coding of the Wavelet Coefficients, Wavelet-based Quadtree coding and Multi-scale Edge detection via CWT); Modulation Domain Analysis – self-study.
3) Image enhancement and image restoration: histogram operators, noise reduction with linear and non-linear filters, unsharp masking, pseudo-colouring, clipping, histogram stretching, image restoration.
4) Image segmentation: Thresholding, Edge detection based on the gradient magnitude (Sobel and Prewitt filters), Edge detection based on the zero crossings of the Laplacian of Gaussian, Canny edge detection, Deformable contours and surfaces, Region based techniques (split-and-merge, watersheds, multi-resolution segmentation - the concept of scale space); Pixel/segment classification (supervised classification by means of linear and quadratic discriminant analysis), unsupervised clustering (e.g. k-means).
5) Mathematical Morphology: general theory for binary and gray value images, examples of operators (erosion, dilation, opening and closing), reconstruction filters, top-hat and bottom-hat filters.
Course notes of the individual sections can be obtained from: ftp://ftp.etro.vub.ac.be/
Complementary study material:
- Digital Picture Processing (2nd Ed.), A. Rosenfeld and A. Kak, Vol. 1 and 2, 1982
- Fundamentals of Digital Image Processing, A. Jain, Prentice Hall, 1989
- Digital Image Processing (3rd Ed.), R. Gonzalez, Addison and Wesley, 1992
- The World according to Wavelets, Barabara Burke Hubbard,A.K. Peters, Wellesley, Massachussets, 1998, ISBN 1-56881-072-5 5
- High performance compression of visual information - a tutorial, Olivier Egger, Pascal Fleury, Touradj Ebrahimi, Murat Kunt, Review Part I: Still Pictures", Proc. IEEE, Vol. 87, No.6, June 1999
- Wavelets and subband coding, Martin Vetterli, Jelena Kovacevic, Prentice Hall, ISBN: 0130970808, 1995
- A wavelet tour of signal processing, S.Mallat, Academic Press, ISBN: 012466606X, 1998.
- The course introduces image representation principles and digital image processing algorithms, including image transforms, image enhancement and restoration, edge detection, image segmentation and image compression. The course describes generic techniques that find their application in a variety of fields, such as visual inspection, medical imaging, compression and transmission of images and video, multimedia applications, machine vision and remote sensing.
- With this course, the student acquires the necessary skills and gathers an in-depth theoretical and practical knowledge up to a stage that he/she should be able to solve various image processing problems.
The final grade is composed based on the following categories:
Oral Exam determines 50% of the final mark.
Written Exam determines 50% of the final mark.
Within the Oral Exam category, the following assignments need to be completed:
Within the Written Exam category, the following assignments need to be completed:
This offer is part of the following study plans:
Master of Applied Sciences and Engineering: Applied Computer Science: Standaard traject (only offered in Dutch)
Master in Applied Sciences and Engineering: Applied Computer Science: Standaard traject