Imperial College London

ProfessorPier LuigiDragotti

Faculty of EngineeringDepartment of Electrical and Electronic Engineering

Professor of Signal Processing
 
 
 
//

Contact

 

+44 (0)20 7594 6192p.dragotti

 
 
//

Location

 

802Electrical EngineeringSouth Kensington Campus

//

Summary

 

Overview

sAMPLING AND rECONSTRUCTION dRIVEN BY SPARSITY MODELS WITH APPLICATIONS TO SENSOR NETWORKS AND NEUROSCIENCE (rECOSAMP)

 

The problem of reconstructing or estimating partially observed or sampled signals is an important one that finds application in many areas of signal processing and communications. Traditional acquisition and reconstruction approaches are heavily influences by classical Shannon sampling theory which gives an exact sampling and interpolation formula for bandlimited signals. Recently, the emerging theory of sparse sampling has challenged the way we think about signal acquisition and has demonstrated that, by using more sophisticated signal models, it is possible to break away from the need to sample signals at the Nyquist rate. The insight that sub-Nyquist sampling can, under some circumstances, allow perfect reconstruction is revolutionizing signal processing, communications and inverse problems. Given the ubiquity of the sampling process, the implications of these new research developments are far reaching.

This project is based on the applicant’s recent work on the sampling of sparse continuous-time signals and aims to extend the existing theory to include more general signal models that are closer to the physical characteristics of real data, to explore new domains where sparsity and sampling can be effectively used and to provide a set of new fast algorithms with clear and predictable performance. As part of this work, he will also consider timely important problems such as the localization of diffusive sources in sensor networks and the analysis of neuronal signals of the brain. He will, for the first time, pose these as sparse sampling problems and in this way he expects to develop technologies with a step change in performance. (More details here)

Reverse Engineering of Audio-Visual Content Data (Rewind)

With the rapid proliferation of inexpensive acquisition and storage devices, multimedia objects can be easily created, stored, transmitted, modified and tampered with by anyone. During its lifetime, a digital object might go through several processing stages, including multiple analog-to-digital (A/D) and digital-to-analog (D/A) conversions, coding and decoding, transmission, editing (either aimed at enhancing the quality, creating new content, mixing pre-existing material, or tampering with the content). 

The REWIND project (funded within the FP7 ICT FET-Open  funding scheme) starts from the fact that each of these processing steps necessarily leaves a characteristic footprint, which can be potentially detected and analyzed to trace back the past history of the available multimedia object in a blind fashion, i.e. without having access to the original content. (More details)

plenoptic signal processing

The availability of multiple views of a scene enables numerous new and exciting applications ranging from 3D and free viewpoint television to robust scene interpretation and object tracking. However, there are still many open challenging issues in terms of processing primarily due to the shear amount of data involved when the number of cameras becomes very large. It is therefore a primordial point to understand how the information is structured and how to take advantage of the inherent redundancy when the cameras are looking at the same scene.

The goal is to understand the nature of the data in multi-view imaging systems particularly in terms of structure and coherence. Specifically, we study the extraction of coherent regions and occlusion boundaries which is an important issue in numerous multi-view image processing applications such as synthesis of novel views, compression and scene understanding. (More details here)

Collaborators

Prof. A. Hirabayashi (visitor in my group), Yamaguchi University, Sampling Theory, 2009 - 2010

Dr M. Tagliasacchi (academic visitor in my lab), Polytechnic of Milan, Image processing, 2012 - 2012

Prof. Thierry Blu, Chinese University of Hong Kong, 2008 - 2013

Guest Lectures

Sampling and Reconstruction driven by sparsity models: Theory and Applications, IEICE, Japan, 2012

Approximate Strang-Fix: Sparse Sampling with any Acquisition Device, Summer Research Institute, EPFL, Lausanne, Switzerland, 2013

Sparse Sampling: Theory and Applications, Keynote talk,, Workshop on Sparsity and Nonlinear Diffusion for Signal and Image Processing,, International Centre for Mathematical Sciences, Edinburgh, November 2009., 2009

On the Sampling and Compression of the Plenoptic Function’, Plenary talk,, IEEE Workshop on Multimedia Signal Processing (MMSP), October 2010, Saint Malo, France, 2010

Sparse Sampling: Sensing Brain Activity at Infinite Resolution’, Invited Speaker, CNS 2013 Workshop on Methods of Information Theory in Computational Neuroscience, Paris, France, 2013

Research Staff

Muammar,H

Visentini Scarzanella,M

Zhang,Y