COMP3072 - Image Processing
This database contains the 2017-18 versions of syllabuses. Syllabuses from the 2016-17 session are available here.
Note: Whilst every effort is made to keep the syllabus and assessment records correct, the precise details must be checked with the lecturer(s).
|Code||COMP3072 (Also taught as: COMPGV12 Image Processing)|
|Prerequisites||Successful completion of years 1 and 2 of the Computer Science, Mathematics and Computer Science or other Physical Science or Engineering programme with sufficient mathematical and programming content.|
|Taught By||Lourdes Agapito (100%)|
|Aims||The first half of this course introduces the digital image, describes the main characteristics of monochrome digital images, how they are represented and how they differ from graphics objects. It covers basic algorithms for image manipulation, characterisation, segmentation and feature extraction in direct space. The second half of the course proceeds to a more formal treatment of image filtering with some indication of the role and implications of Fourier space, and more advanced characterisation and feature detection techniques such as edge and corner detection, together with multiresolution methods, treatment of colour images and template matching techniques. The course allows students to explore a range of practical techniques, by developing their own simple processing functions either in a language such as Java and/or by using library facilities and tools such as MatLab or IDL.|
|Learning Outcomes||To understand (i.e., be able to describe, analyse and reason about) how digital images are represented, manipulated, encoded and processed, with emphasis on algorithm design, implementation and performance evaluation.|
Introduction to the digital image
Why digital images?
The (film and) digital camera.
Data types and 2d representation of digital images.
Characteristics of grey-level digital images
Discrete sampling model.
Thresholding and thresholding algorithms.
Performance evaluation and ROC analysis.
Connected components labelling.
Region growing and region adjacency graph (RAG).
Split and merge algorithms.
Grey level transformations.
Erode and dilate as max and min operators on binary images.
Open, close, thinning and other transforms.
Medial axis transform.
Introduction to grey-level morphology.
Linear and non-linear filtering operations.
Sub-sampling and interpolation as convolution operations.
Calculation of region properties.
Boundary coding line descriptors from boundary coding and from moments.
Image search and multi-resolution algorithms.
EEdge and corner detection
Edge enhancement by differentiation.
Effect of noise, edge detection and Canny implementation.
Edge detector performance evaluation.
Image structure tensor.
Relationship to image auto-correlation.
Characterisation and Harris corner detector.
Representations of colour in digital images.
Pixel-wise (point) operations.
Colour invariants and Finlayson colour constancy algorithm.
TeTemplate matching and advanced topics
Similarity and dissimilarity matching metrics.
L2 metric and relationship to cross-correlation2D object detection, recognition, location.
Sub-pixel accuracy and performance evaluation.
Method of Instruction
Lecture presentations with associated class coursework and laboratory sessions. There are 4 pieces of coursework, all weighted equally.
The course has the following assessment components:
- Written Examination (2.5 hours, 80%);
- Coursework Section (3 pieces, 20%).
To pass this course, students must:
- Obtain an overall pass mark of 40% for all componenets combined;
- Obtain a minimum mark of 30% in each component worth ≥ 30% of the module as a whole.
Reading list available via the UCL Library catalogue.