EngD (Engineering Doctorate) - Signal and Image Processing for very large format imagery

Project available to UK applicants only.

The EngD is an alternative to a traditional PhD aimed at students wanting a career in industry. Students spend about 75% of their time working directly with a company in addition to receiving advanced-level training from a broad portfolio of technical and business courses. On completion students are awarded the PhD-equivalent EngD.

Industrial Partners: Thales

Project Description:

The technical challenges involved in capturing, storing and processing very large format airborne imagery are of current topical interest especially in the defence industry. The additional dimension introduced by multi and hyper spectral data make these challenges even more acute. Recent initiatives in BIG DATA serve to highlight the importance of this issue.

The ability to identify specific image features and then to rapidly search large image data sets to pick out equivalent features is one with applications in many fields.

This project will investigate the technical challenges involved in handling large format imaging and develop software tools designed to rapidly locate specific features. Ideally the data should be capable of being searched for features whilst in its compressed state.

The sheer volume of data generated by hyperspectral imaging make compression inevitable. Techniques which capture characteristics in both the spatial and spectral domain and exploit their mutual redundancy will be most able to achieve efficient compression in terms of data reduction, speed of compression/ decompression as well as usability of the data in its compressed format.

The novel aspects will be a suite of the search and classification tools for finding features indicated by examples in large complex data.
An operator would indicate a feature in a data set and the tool would extract combinations of spatial and spectral features and determine similar sets of features within the data. It is a type of google goggles but for HSI data.
Ideally working on the compressed format.

The benefit to Thales is in having a robust search tool for marking targets such as vehicles then being able to search large data sets for similar targets.
It also give access to the expertise in the University research group in image processing, feature extraction and novel classifiers as well as a new potential high skilled employee.

The candidate will have excellent knowledge of mathematics, computer programming (both matlab and C++), and signal/image processing plus a pro-active approach to their work. They will be involved in design and execution of practical experiments for data capture as well developing algorithms for processing and compression

The skills acquired during this project will be more generally applicable in digital imaging and processing of large data sets. A potential commercial application of crop health monitoring is an example of non military use of these search tools.
The tools would be trained to predict outbreaks of crop blight or infestation so that corrective action could be taken early to treat or isolate the outbreak.