Computer Vision is concerned with the automatic interpretation and analysis of images. The goal is to enable computers to understand visual information in a similar way to humans. Such an ability is fundamental for solving many problems in areas such as industrial inspection, medical image analysis, robot navigation, biometrics, surveillance and security. This module aims to provide a comprehensive introduction to the subject, covering the basic concepts, methodologies and tools of image analysis and interpretation. This module will also explore the biological visual system, which has solved the problem of vision far better than contemporary computer vision systems. This will highlight the limitations of computer vision and suggest approaches for building improved, biologically-inspired, vision systems.
At the end of the course, students should have achieved:
An understanding of common computer vision algorithms and methods. The ability to implement methods for solving simple image analysis problems. The ability to apply the relevant mathematics that underlie computer vision. An understanding of mechanisms for edge detection, image segmentation, and object recognition. An understanding of the human visual system and theories of the neural mechanisms that underlie visual cognition.
Image formation: the physics of image formation, the geometry of image formation, image coding and representation, the human visual system.
Low-level vision: image processing (filtering, convolution), feature detection (edges, corners), representations in V1 and V2.
Mid-level vision: grouping and segmentation, stereo and depth, motion and video, Gestalt principles, lateral connectivity in V1 and V2.
High-level vision: object categorization and recognition.
Module assessment - more information