Department of Computer Science, Unit Catalogue 2008/09
CM30080 Computer vision
| Credits: 6 |
| Semester: 2|
|Assessment: CW 25%, EX 75%|
|Before taking this unit you must (take CM10134 or take CM10140) or (take CM20001 and take CM20144) or take MA20010|
or equivalents (authorised by the Director of Studies).
Aims: To understand the ways of analysing images to get information out of them. In order to achieve this it will be necessary to understand the underlyin mathematics and computer techniques.
Students will be able to:
1. Distinguish low-level from high-level Computer Vision methods, and appreciate the vision problem;
2. Describe edge detection as a linear filter and distinguish between linear filtering and morphology;
3. Describe multi-camera geometry and understand its value in applications such as mosaicing and reconstruction;
4. Understand texture and segmentation, and the role of high-level models in recognition;
5. Appreciate a broad range of contemporary Computer Vision.
Application of number (T/F).
Low level vision: Convolution and linear filtering, edge detection and blurring; the role of scale. Morphology. Texture descriptors. Multi-camera vision: Homographies, epipolar geometry, and the fundamental matrix. Mosaicing and 3D reconstruction. Segmentation: Hough transforms, unsupervised clustering, scale sieves. Recognition: The role of prior models: templates, geometry, and statistics.