Multidimensional Radar Imaging for Target Identification

FOREWORD

The SET-273 Research Specialists’ Meeting (RSM) on “Radar Imagining for Target Identification” was convened from 25-26 October in Marseille France. Over 86 participants from 14 countries exchanged results of previous SET-250 Multi-Dimensional Radar Collections from 4 radars that had multiple channels and multi-polarization of synthetic aperture radar (SAR) targets. In addition, there were presentations on emerging technologies, closely related to the SM interests.

There is an increasing awareness of that using of multi-channel, multi-static, multi-frequency and multi-polarization radar imaging can greatly improve target classification performance. Small drones are being proliferated widely by asymmetric warfare agents. And these threats are very difficult to discriminate from birds and other clutter over the sea or within proximity to land cultural objects. The development of efficient, real-time Machine Learning algorithms have provided near real-time discrimination of tactical targets . However, there is insufficient images to train and apply these Machine Learning algorithms. As a result, the researchers have outlined new modeling and image synthesis to complement the available images. In addition, the use of bistatic and multiple-dimension imaging can provide a significant spatial awareness of targets that are obscured by forests and buildings.

Author(s): Milan Rozel; Philippe Brouard; Hélène Oriot  

DOI: 10.14339/STO-SET-273-06     |     

The last decade has seen great developments in the domain of small unmanned aerial vehicles (UAVs). Today UAVs are able to carry out a wide range of missions and recent developments in conflict zones, involving states and non-state actors, have shown how effective they can be. This effectiveness lies in part on their ability to foil conventional surveillance systems, as low observable targets

Author(s): Luc Vignaud;  

DOI: 10.14339/STO-SET-273-19     |     ISSN: TBD  

I present IGAN (Inferent Generative Adversarial Networks), a neural architecture that learns both a generative and an inference model on a complex high dimensional data distribution, i.e. a bidirectional mapping between data samples and a simpler low-dimensional latent space. It extends the traditional GAN framework with inference by rewriting the adversarial strategy in both the image and the latent space with an entangled game between data-latent encoded posteriors and priors

Author(s): Alessandro Vetere; Eugenio Pino; Adriano Meta  

DOI: 10.14339/STO-SET-273-21     |     ISSN: TBD  

Automatic Target Recognition (ATR) with Artificial Intelligence (AI) plays a key role in modern surveillance and reconnaissance activities. The enabling technology for ATR is the availability of a database that is representative of realistic situations including targets and backgrounds. When applied to radar imaging the problem of collecting enough data of targets under different configurations and conditions becomes extremely complicated and unaffordable

Author(s): Juha Jylhä; Minna Väilä; Marja Ruotsalainen; Juho Uotila  

DOI: 10.14339/STO-SET-273-03     |     ISSN: TBD  

Synthetic Aperture Radar (SAR) imaging using a small and lightweight aircraft, such as a drone, sets special requirements on the image formation. Such an aircraft is unable to maintain strictly straight and level flight as it is easily susceptible to the small effects of the atmosphere, such as wind. With conventional SAR methods, a highly fluctuating flight trajectory leads to compromises in image focusing causing distortions in the image

Author(s): Patrick Berens; Michael Caris;  Ingo Walterscheid  

DOI: 10.14339/STO-SET-273-07     |     ISSN: TBD  

During a measurement campaign of the NATO research group SET-250 SAR measurements have been performed with two SAR sensors of Fraunhofer FHR using different frequency bands: the MIRANDA-94 and the PAMIR-Ka. We processed the data to create SAR images with approximately 10 cm resolution in both range and cross range direction and analysed the results

Author(s): Daniel Andre; Richard Sabiers; Mark Finnis  

DOI: 10.14339/STO-SET-273-05     |     ISSN: TBD  

Through-Wall (TW) Synthetic Aperture Radar (SAR) imagery can be difficult to interpret due to several factors including signal attenuation in highly cluttered environments, target overlay, difficult to interpret low SAR resolution and low frequency scattering responses. One approach which may help improve image interpretability is to employ 2D SAR apertures with multiple distributed receivers in all polarizations