|
|
|
Moeness Amin (Villanova University, USA)
This talks considers nonstationary signal analysis in view of the signal sparsity properties. We examine these signals, which arise in numerous applications, within the framework of compressive sensing (CS) and sparse reconstructions. We present two general approaches to incorporate sparsity into time-frequency signal presentation (TFSR). In the first approach, quadratic TF distributions (QTFDs) are derived based on optimal multi-task kernel design. In this case, sparseness in the TF domain presents itself as a new design task, adding to those related auto-term preservation and cross-term suppression. In contrast to QTFDs, we also provide a second approach for signal TF signature estimation where sparse reconstruction is used in lieu of direct Fourier transform that maps the signal from the time or one joint-variable domain to another. It is shown that multiple measurement vector methods and block sparsity techniques play a clear and fundamental role in improving signal local power representations. Examples of both approaches are provided. Analysis is supported by simulations with synthesized data, experiments with real Doppler and microDoppler data measurements of radar returns associated with human motions, and by electromagnetic modeled.
Holger Rauhut (RWTH Aachen University, Germany)
This talk intends to give an introduction and overview on the mathematical foundations of compressive sensing with particular focus on applications in radar.
The key ingredients of compressive sensing are sparsity, efficient algorithms and random matrices. This combination allows to show rigorous recovery guarantees for
reconstruction algorithms (such as l1-minimization) and measurement schemes. In most applications including many radar scenarios the matrix modeling the measurement
process obeys some structure while compressive sensing guarantees are usually based on random constructions. This leads to the study of various types of
structured random matrices. Randomness may in practice be realized by chosing the probing signal, the antenna locations or both at random. The talk will discuss
several of such measurement matrices and available recovery guarantees. In particular, we will cover discuss partial random circulant matrices,
time-frequency structured random matrices, measurements with random antenna arrays as well as MIMO radar with randomly chosen probing signals.
Müjdat Cetin (Sabanci University, Turkey)
In this talk we first present an overview of several lines of inquiry in which we have been involved and that lie at the intersection of two domains: sparse signal representation and synthetic aperture radar (SAR) image formation. This historical overview contains (i) analysis and synthesis-based sparse signal representation formulations for SAR image formation together with the associated imaging results; (ii) sparsity-based methods for wide-angle SAR imaging and anisotropy characterization; (iii) sparsity-based methods for joint imaging and autofocusing from data with phase errors; (iv) techniques for exploiting sparsity for SAR imaging of scenes containing moving objects, and (v) compressed sensing-based analysis and design of SAR sensing missions. Next, we describe more recent work together with our perspective for future research in this domain. Specific points of discussion and challenges posed will include reducing computational complexity and performing distributed processing, establishing stronger connections between imaging and decision-making, using effective machine learning ideas to tune the notion of sparsity to a particular context, and going beyond sparsity to exploit other forms of simple structures present in the data as well.
Mujdat Cetin is an Associate Professor at Sabanci University, Istanbul, where he currently directs the Signal Processing and Information Systems Laboratory (http://spis.sabanciuniv.edu). From 2001 to 2005, he was with the Laboratory for Information and Decision Systems, MIT. He received the Ph.D. degree from Boston University in 2001. Dr. Cetin has held visiting faculty positions at MIT, Northeastern Univ., and Boston Univ. Dr. Cetin's research interests lie within the field of statistical signal and image processing, and include image reconstruction and restoration, radar imaging, sparse signal and image representation, biomedical image analysis, shape-based image segmentation, brain-computer interfaces, machine learning, data fusion, sensor array signal processing, and inference in sensor networks. His publications in these areas have received more than 3500 citations so far based on Google Scholar records. Dr. Cetin has received several awards including the 2010 IEEE Signal Processing Society Best Paper Award; the 2007 EURASIP/Elsevier Signal Processing Best Paper Award; the 2013 IET Radar, Sonar and Navigation Premium (Best Paper) Award; the 2008 Turkish Academy of Sciences Distinguished Young Scientist Award; the 2010 METU Mustafa Parlar Foundation Research Incentive Award; and the 2006 TUBITAK Career Award, as well as four conference best paper awards. Dr. Cetin was the Technical Program Co-chair for the International Conference on Information Fusion (FUSION) in 2013; for the International Conference on Pattern Recognition (ICPR) in 2010; and for the IEEE Turkish Conference on Signal Processing, Communications and their Applications in 2006. Dr. Cetin is currently an Associate Editor for the IEEE Transactions on Image Processing. He previously served as an Associate Editor for the IEEE Signal Processing Letters and the IEEE Transactions on Cybernetics; as a Guest Editor for Pattern Recognition Letters; and as an Area Editor for the Journal of Advances in Information Fusion. He is an Associate Member of the IEEE Bioimaging and Signal Processing Technical Committee, and previously served as a member of the International Association for Pattern Recognition (IAPR) Conferences and Meetings Committee, as well as a member of the IAPR ICPR Liaison Committee. Dr. Cetin also serves as a EURASIP Liaison Officer for Turkey and a best paper award jury member for the EURASIP Journal on Advances in Signal Processing. |