University of Leeds Featured PhD Programmes
University of Kent Featured PhD Programmes
University of Liverpool Featured PhD Programmes
King’s College London Featured PhD Programmes
Birkbeck, University of London Featured PhD Programmes

AI and physics-based approaches to bounding visual uncertainty (FINLAYSONG2U18SCI50)

This project is no longer listed on FindAPhD.com and may not be available.

Click here to search FindAPhD.com for PhD studentship opportunities
  • Full or part time
    Prof G D Finlayson
    Dr M Mackiewicz
  • Application Deadline
    No more applications being accepted
  • Funded PhD Project (European/UK Students Only)
    Funded PhD Project (European/UK Students Only)

Project Description


The images recorded by a camera depend on 3 things [i]: the colour of the objects in the scene, the viewing conditions (e.g. colour of the lights) and the physical characteristics of the camera. Even when the camera and viewing conditions are well understood – for example, we take a picture under lab conditions – there is considerable uncertainty in determining whether one object is made of the same material as another (in detail, whether the underlying spectral reflectances are the same). This uncertainty increases outside the lab.

In this PhD project we aim to develop new algorithms for bounding the uncertainty in camera measurements. Specifically, we wish to – based on camera images – determine what materials an object is not made from. Knowing the answer to this question is useful in many applications. For example, an automotive vision system may wish to determine whether a particular region in the image is or is not a road sign (and if it is a road sign then it would deploy more computational resources to interpreting the sign). The project will begin by looking – and advancing – the classical methods for material determination that depend on inverting the physics of image formation [ii,iii]. While these methods are ‘elegant’ they are computationally inefficient, so the material determination problem will be reformulated in an AI framework, e.g. [iv] (where we will aim to provide an improved classification rate and a much quicker overall solution to the problem). Depending on the student’s interest and background, the project may focus more on the classical or AI solutions.

The project runs in the Colour & Imaging lab at the University of East Anglia in collaboration with Apple Inc.

For more infomation on the supervisor for this project please go here: https://www.uea.ac.uk/computing/people/profile/g-finlayson

Type of programme:PhD

Start date of the project: October 2018

Mode of study: Full Time

Acceptable first degree Computer Science, Physics, Mathematics, Engineering or other numerate discipline

The standard minimum entry requirement is 2:1


Funding Notes

This PhD studentship is jointly funded for three years by Faculty of Science and Apple. Funding comprises home/EU fees, an annual stipend of £14,296 and £1000 per annum to support research training. Overseas applicants may apply but are required to fund the difference between home/EU and overseas tuition fees (in 2018/19 the difference will be at least £10,905 for the School of Computing Sciences but fees for 2018/19 are yet to be confirmed and are subject to an annual increase).

References

i) Vrhel M. et al, “Color image generation and display technologies,” IEEE Signal Processing Magazine, 2005.

ii) Finlayson G. and Morovic P., “Metamer Sets,” Journal of the Optical Society of America A, 2005

iii) Morovic P. and Finlayson G., “Metamer Set Based Approach to estimating surface reflectance from camera RGB,” Journal of the Optical Society of America A, 2006.

iv) Peng. Et al, “Large Kernel Matters -- Improve Semantic Segmentation by Global Convolutional Network”, IEEE International Conference on Computer Vision and Pattern Recognition, 2017

Related Subjects



FindAPhD. Copyright 2005-2019
All rights reserved.