Hensoldt (Germany)

FrodoKEM is a post-quantum Key Encapsulation Mechanism (KEM) selected by the Bundesamt für der Sicherheit in der Informationstechnik (BSI) and the Agence Nationale de la Sécurité des Systémes d’Information (ANSSI). Although its theoretical security is still intact to this day, the FrodoKEM sensitive message can be recovered through other means, such as Side-Channel Analysis (SCA). In this work, we expand on a previous attack by Ravi et al. by detailling its effect on the different security levels of FrodoKEM. We study several strategies to recover its message in a minimal number of chosen ciphertexts queries.

We present an overview of the Enhanced Multi-Source Information Fusion(EMSIF)projectas well as the research work carried out during the project. The aim is to create an improved situation picture for the future application domain Advanced Air Mobility. The improvement comprises two parts On the one hand new types of data sources are integrated, such as OSINT data and Background Knowledge(BK)On the other hand conventional Multi-Source Information Fusion (MSIF)is often limited to the initial stage which focuses on estimating the states of individual entities. In contrast. the subsequent stage focuses on a higher level of inference which takes into account the relations between these entities to identify critical situations and derive conclusionsThat allows the operator to detect and evaluate hazardous situations more easily, even in crowded environments

Marczali Henrik (Marcali, 1856 – Budapest, 1940) volt az első professzionális történetíró Magyarországon. 1875–1877-ben Berlinben Theodor Mommsen, Wilhelm Wattenbach, Karl Wilhelm Nitzsch és Georg Waitz, Párizsban Jules Quicherat és Gabriel Monod, Oxfordban William Stubbs tanítványaként megismerte és elsajátította a korabeli legmagasabb színvonalú szakszerű forráskritikai szemléletét, módszerét, technikáját és az alkalmazásukhoz szükséges készségeket. George Peabody Gooch cambridge-i történész 1913-ban úgy értékelte, hogy Marczali monográfiáival született meg Magyarországon a provincializmus kötelékeiből kiszabaduló tudományos történetírás. Történetírói oeuvre-je és tanári hivatása szorosan illeszkedtek egymáshoz. Ő volt az, aki a történész szakképzést meghonosította a Budapesti Tudományegyetemen és a történelemtanárok és kutatók egész nemzedékét képezte ki a tanárképző intézetben és az általa megtervezett és létrehozott, berlini mintájú történeti szemináriumban. A következő nemzedék Budapesten képzett történészei és történelem tanárai valamennyien az ő tanítványai voltak. A tanulmány történetírói előfeltevései, kulcsfogalmai (műveltség, civilizáció, erkölcs), nemzetértelmezése és a prófétai örökség alkalmazása rekonstrukcióján és értelmezésén keresztül közelíti meg értelmiségi szerepfelfogását. Perfekcionista felfogásának kialakulásában megalapozó szerepet játszott édesapja, Morgenstern/Marczali Mihály rabbi és mentora, Kármán Mór értelmezése a zsidó hagyományról, a prófétai örökségről, amelyet Marczali átvett és alkalmazott.

Simulated remote sensing images bear great potential for many applications in the field of Earth observation. They can be used as controlled testbed for the development of signal and image processing algorithms or can provide a means to get an impression of the potential of new sensor concepts. With the rise of deep learning, the synthesis of artificial remote sensing images by means of deep neural networks has become a hot research topic. While the generation of optical data is relatively straightforward, as it can rely on the use of established models from the computer vision community, the generation of synthetic aperture radar (SAR) data until now is still largely restricted to intensity images since the processing of complex-valued numbers by conventional neural networks poses significant challenges. With this work, we propose to circumvent these challenges by decomposing SAR interferograms into real-valued components. These components are then simultaneously synthesized by different branches of a multi-branch encoder–decoder network architecture. In the end, these real-valued components can be combined again into the final, complex-valued interferogram. Moreover, the effect of speckle and interferometric phase noise is replicated and applied to the synthesized interferometric data. Experimental results on both medium-resolution C-band repeat-pass SAR data and high-resolution X-band single-pass SAR data, demonstrate the general feasibility of the approach. • Complex-valued SAR interferograms are simulated with deep learning. • Complex-valued target data is split into real-valued components. • Multitask learning is used to synthesize real-valued components. • Approach combines physical model and deep learning.

A fast synthesis method for shaped beams in arrays that exploits the sensitivity of the radiation pattern to element excitations is presented. The excitation phases are determined such that the array radiates a pattern that optimally approaches a predefined template. To efficiently achieve this, an iterative procedure making use of a freely accessible auto-differentiation framework is presented. While the conventional array factor remains strictly valid for infinite arrays without edge reflections, its application to finite arrays may introduce significant errors due to edge truncation. To address this challenge, we enhance the array factor with a correction term derived herein. We evaluate the prediction accuracy of our method using a challenging power template covering a wide dynamic range across the array's field of view. Additionally, we validate our approach through experimental implementations. These experiments demonstrate the efficacy and robustness of our synthesis method.

In this report we describe the effect of the uncoated and the coated grinded lens edges on the imaging performance of a high performance short wave infrared (SWIR) imager inside a fully achromatized multispectral SWIR/VIS zoom camera. The experimental result of various black coatings for the lens edges to prevent stray light in the VIS-SWIR spectrum is presented. SWIR cameras show significant advantages for the long range observation under different weather conditions (e.g, haze, mist or fog) due to the reduction of wavelength-dependent scattering extinction. The design of such a SWIR imager in a combined visible (450nm to 900nm) and the SWIR spectral (900nm to 1700 nm) camera with a continuous optical zoom factor of 15x for a limited space and relatively high pixel pitch of 15μm force the use of relatively high F-numbers to achieve narrow fields of view less than 1.5° with a reasonable resolution is a great challenge. Particularly unwanted stray light inside the camera significantly degrades the imaging performance of the SWIR sensor. A good anti-reflection coating on the lens elements improves the ghost images. However, especial prevention techniques are required to reduce the flare stray light that typically arises from reflection or scattering from non-optical path surfaces e.g., mount surfaces or edge surfaces of lenses. The mechanical baffles or stops can reduce these stray lights but it requires precise design and cost. The most common approach to prevent the stray light from the edges of the optical elements is to apply a black coating on the grinded lens edges. The available coatings for edge blackening are typically optimized for VIS spectral range and their blackening effect is limited in the SWIR spectrum by Fresnel reflection due to the index of refraction mismatch at the glass boundary. Here we have applied the appropriate black coating on the selective lens edges based on simulation as well as experiments and demonstrated the improved stray light performance in the SWIR channel.

We present a new approach for the tolerancing of correlated optical systems. These systems consist of at least two optical channels, which are linked together. To represent the correlations e.g., for compensators correctly in the tolerancing procedure, tabulated pseudo random numbers are used. This leads to efficient tolerancing and accurate results.

Visually detecting camouflaged objects is a hard problem for both humans and computer vision algorithms. Strong similarities between object and background appearance make the task significantly more challenging than traditional object detection or segmentation tasks. Current state-of-the-art models use either convolutional neural networks or vision transformers as feature extractors. They are trained in a fully supervised manner and thus need a large amount of labeled training data. In this paper, both self-supervised and frugal learning methods are introduced to the task of Camouflaged Object Detection (COD). The overall goal is to fine-tune two COD reference methods, namely SINet-V2 and HitNet, pre-trained for camouflaged animal detection to the task of camouflaged human detection. Therefore, we use the public dataset CPD1K that contains camouflaged humans in a forest environment. We create a strong baseline using supervised frugal transfer learning for the fine-tuning task. Then, we analyze three pseudo-labeling approaches to perform the fine-tuning task in a self-supervised manner. Our experiments show that we achieve similar performance by pure self-supervision compared to fully supervised frugal learning.

Systematic design of a well corrected broadband refractive lens system is described. We approach the design solution beginning from an appropriate starting point that is the Lister-Petzval form with two positive groups separated by an airspace. By identifying the limiting aberrations at each phase, additional lens elements such as an image side field flattener and a quasi-concentric meniscus lens are added. We also methodically consider the performance improvement with the conversion of the quasi-concentric meniscus into a bi-aspherical element. The impact of the biaspherical element on aberration correction potential of this lens form, along with glass considerations for effective color correction are detailed.