Home

Sponsored by

Delivered by

Home

Classification & Multi-modal Processing Theme

This theme is concerned with broadband signal separation, and detection and classification of co-channel signals transmitted from many targets and received simultaneously. For the scenarios involving rapidly manoeuvring targets, this theme also addresses the non-stationary processing issues.

Theme Meeting

06 Feb 2012 testt

testt

theme

04 May 2011 EEE Building, Imperial College London

The 6th UDRC Theme Meeting and Industry Day took place at Imperial College on 4 May 2011. The main purpose of this event was to provide opportunities to colleagues from industry to understand, interact and collaborate with researchers working on various UDRC projects.  The goal is to exploit the practical use of our research results by the industry.

The project presentations from this event can be found at http://www.mod-udrc.org/announcement/328.
The aim of this project is to design and test measures of context and demonstrate how information about the relationship between all the objects present in high-resolution sonar imagery can be used to detect “unauthorised” objects that are not part of the natural set of objects that are found on the seabed. This is motivated by applications such asminehunting sonar or the detection of unexploded ordnance. Current schemes to detect underwater objects work well in normal sonar conditions when the object is known a priori but are less effective in the presence of clutter and variability or when the “unauthorised” objects are obscured or not already known e.g. when the objects are disguised or improvised.

Project Supervisor

Dr. Duncan Williams

Dr Duncan Williams is the Capability Leader for Acoustics in the Physical Sciences Department at the Defence Science & Technology Laboratory. Previous to joining Dstl he held a two-year post-doctoral appointment at the Marine Physical Laboratory at Scripps Institution of Oceanography, received his PhD in applied mathematics from the University of Nottingham, and studied mathematics at Oxford University. His research interests now cover structural acoustics, acoustic sensing, and signal processing and he is currently a visiting lecturer in the Department of Mathematics at Imperial College London.

theme

15 Dec 2010 Room 503, EEE Building, Imperial College London

This was a joint  “Classification and Multimodal Processing” and “Distributed Signal Processing” Themes meeting at Imperial College London, EEE Building (Room 503) on 15 December 2010.

theme

30 Sep 2010 Room 403, EEE Building, Imperial College London

The purpose of this theme meeting was to give to the UDRC researchers an insight into what problems the military are working on and to get them thinking about these problems and how to tackle them. More specifically, Dstl presented two MOD research challenges and let people self-select into groups and go off to two rooms to discuss approaches, then reconvene and present their solutions back to the audience.

ttesttest

Project Supervisor

Dr. Tania Stathaki

Tania Stathaki was born in Athens, Hellas. In September 1991 she received the Masters degree in Electronics and Computer Engineering from the Department of Electrical and Computer Engineering of the National Technical University of Athens (NTUA) and the Advanced Diploma in Classical Piano Performance from the Orfeion Athens Conservatory of Music. She received the Ph.D. degree in Signal Processing from Imperial College in September 1994. She is currently a Reader in the Department of Electrical and Electronic Engineering of Imperial College. Previously, she was Lecturer in the Department of Information Systems and Computing of Brunel University in UK, Visiting Lecturer in the Electrical Engineering Department of Mahanakorn University in Thailand and Assistant Professor in the Department of Technology Education and Digital Systems of the University of Pireus in Greece. Her current research interests lie in the areas of signal and image processing and computer vision.

Publication Summary

On learning for fusion over fading channels in wireless sensor networks
Jinho Choi and Duc To
In Proc. IEEE ISWPC 2010
2010

Abstract

In order to derive optimal/suboptimal fusion rules,in general, it is assumed that statistical properties of sensors’decisions are known to a fusion center in distributed detectionfor wireless sensor networks. However, if sensors are deployed tounknown environments, these statistical properties may not beavailable in advance and should be estimated by the fusion center.To address this problem, in this paper, we study unsupervisedlearning to estimate the values of the parameters that characterizestatistical properties for wireless sensor networks employinga bandwidth efficient multiple access scheme, e.g., the type-basedmultiple access (TBMA), over Rayleigh fading channels (whichwould be realistic channels when there is no line-of-sight betweensensors and fusion center). Through simulations, we can showthat unsupervised learning can be used in deriving decision rulesat the fusion center from decisions transmitted by sensors overwireless fading channels.

In many applications, events are presented or displayed visually to an operator who is then responsible for detecting the presence and identity of threat targets using this visual information. This is not always effective for the detection of transient events. Such events are more likely to be detected by an auditory display and operators typically rely on listening to make a decision. This is mostly due to the human auditory system excelling in the detection of transient sounds in the presence of noise and the advantage of combined auditory and visual processing. Notwithstanding this superiority, there is still no effective way to automate this integration of auditory and visual information as part of the system display. Routine experience is that sonar post event analysis detects and classifies targets that were not reported operationally. The aim of this task is to identify candidate signal processing techniques for automated transient detection that exploits combined auditory and visual processing. The emphasis is on ways to integrate the auditory and visual information that characterise transient events. The task will compare early and late integration methods for auditory-visual processing.

Project Supervisor

Mr. Adrian Brown

Adrian Brown BSc MSc is a Principal Scientist in Dstl. His initial research was in ocean measurement and modelling with a particular interest in oceanographic internal waves. This work involved several sea trials including a month in the Atlantic aboard a weather ship. He then spent some time working in Operational Analysis (aka Operational Research) with a focus on engagement simulation modelling of submarine detection. Subsequently he was responsible for the MoD's reference document on Sonar Modelling, leading a team working on sonar performance modelling. More recently he has acted as scientific advisor to a major submarine sonar procurement, which included planning of and participation in submarine sonar trials.

theme

19 May 2010 Room 503, EEE, Imperial College London

The 3rd "Classification and Multi-modal Processing" Theme Meeting took place at Imperial College on 19 May 2010

Coming soon

Project Supervisor

Dr. Jonathan Perry

Jon Perry was born in Bristol, UK, in 1962. He studied Physics at Imperial College (BSc 1983, PhD 1988). He joined Space Department in what was then known as the Royal Aircraft Establishment as a Senior Research Fellow in 1986 and has continued to work in the scientific branch of the Civil Service ever since. Jon is now a Principal Scientist in Sensors and Countermeasures Department of Dstl (the Defence Science and Technology Laboratory – RAE’s successor organisation). His research interest is the imaging by aircraft and spaceborne reconnaissance radar of a range of ocean surface features, such as surface expressions of bathymetry and of internal waves, surface waves and wave wakes.