Please note: there may be some adjustments to the teaching arrangements published in the course catalogue for 2020-21. Given current circumstances related to the Covid-19 pandemic it is anticipated that some usual arrangements for teaching on campus will be modified to ensure the safety and wellbeing of students and staff on campus; further adjustments may also be necessary, or beneficial, during the course of the academic year as national requirements relating to management of the pandemic are revised.

MSc SIS Edinburgh Course - Image and Vision Computing PHYS5087

  • Academic Session: 2021-22
  • School: School of Physics and Astronomy
  • Credits: 10
  • Level: Level 5 (SCQF level 11)
  • Typically Offered: Semester 2
  • Available to Visiting Students: No
  • Available to Erasmus Students: No

Short Description

The course proceeds in five parts from foundational concepts such as image formation, through to low-level image processing operations, before building upon those to develop image representations, and use those representations for higher-level tasks such as recognition and detection. The following four parts are roughly two weeks each.

Image formation: the basic mathematics and physics of how images are formed based on light reflected by real-world objects. Includes ideal pinhole camera and lens models. Some basic 3D geometry, radiometry and photometry.

Low level image analysis: We will introduce basic algorithms such as convolution and filtering for image processing, and RANSAC for fitting. These will be applied for tasks such as edge detection, and line-fitting. To provide a taste of recognition students will perform shape recognition using Bayes theorem.

Image Representations: To support working with more unconstrained realistic images, we next introduce feature representations for both local and global features including colour histograms, HOS/SIFT, and descriptor bag of words.

High-level imagine analysis: Building upon these image representations, we discuss the topical tasks of object recognition and sliding window-based object detection.

Applications: Finally, we finish up with an introduction to some applications including basic video processing (optical flow), and foreground detection.

Timetable

None

Requirements of Entry

None

Excluded Courses

None

Co-requisites

PHYS5044 Fundamentals of Sensing

Assessment

None.

Main Assessment In: April/May

Are reassessment opportunities available for all summative assessments? No

Edinburgh University does not provide resit examinations for MSc students. In cases where an assessment is affect by reasons of good cause a revised mark, based on other completed assessments, may be substituted.

Course Aims

The aims of this course are to analyse how images are formed given objects in the three dimensional world, and the basics of how computer vision inverts this process - computing properties of the world from digital images. The course will discuss topics including basic image formation, image processing, detection, matching and recognition that allow computers to understand the world based on image content.

Intended Learning Outcomes of Course

By the end of this course students will be able to:

1. Explain the basic physics and mathematical principles of image formation.

2. Demonstrate a detailed knowledge of basic image processing operations such as convolution.

3. Write programs to solve basic image analysis tasks such as edge detection and line fitting.

4. Explain the concepts of local and global image descriptors, and descriptor matching.

5. Write programs to perform image analysis tasks of recognition and detection.

Minimum Requirement for Award of Credits

Students must submit at least 75% by weight of the components (including examinations) of the course's summative assessment.