Automatic Classification of Targets in High Resolution Radar Imagery

The research will be focussed on the development of new advances in machine learning to improve the ability of the classification system to distinguish between different objects. A good classification system has the potential to greatly ease the operator’s workload in a maritime patrol aircraft, particularly when there are a large number of ship targets in the surveillance region. The Defence Science and Technology Group (DST Group) has an existing research program in this area, and is looking for ways to improve the performance of their system. Potential research directions include:

• Machine learning techniques with relatively small amounts of data,
• Matching of radar imagery against templates derived from CAD models Methods for combining different classifiers or their outputs.
• Use of metadata or contextual information.

Support: By working with DST Group, there will be access to real ship imagery to develop ideas and a copy of the existing classifier software as a baseline to evaluate the performance of any newly developed classification schemes.

We offer a $5K top-up for 3 years plus up to $10k for travel support. Applicant would need to be an Australian citizen.

Enquiries: 
Dr Brian Ng
School of Electrical & Electronic Engineering
Tel (08) 8313 5054