This PDF file contains the front matter associated with SPIE Proceedings Volume 13199, including the Title Page, Copyright information, Table of Contents, and Conference Committee information.

Deep Learning based architectures such as Convolutional Neural Networks (CNNs) have become quite efficient in recent years at detecting camouflaged objects that would be easily overlooked by a human observer. Consequently, countermeasures have been developed in the form of adversarial attack patterns which can confuse CNNs by causing false classifications while maintaining the original camouflage properties in the visible spectrum. In this paper, we describe the various steps in generating suitable adversarial camouflage patterns based on the Dual Attribute Adversarial Camouflage (DAAC) technique for evading the detection by artificial intelligence as well as human observers which was proposed in [Wang et al. 2021]. The aim here is to develop an efficient camouflage with the added ability to confuse more than a single network without compromising camouflage against human observers. In order to achieve this, two different approaches are suggested and the results of first tests are presented.

Due to the enormous development in the field of artificial intelligence, especially in the area of reconnaissance, detection and recognition, it has become absolutely necessary to think about methods of concealing one's own military units from this new threat. This publication aims to provide an overview of counter ai approaches against enemy reconnaissance, and the possibilities to assess the effectiveness of these methods. It will focus on explainable AI and the camouflaging of key features as well as the possibility of dual attribute adversarial attack camouflage. These are mathematically optimised patterns that drive an AI-based classifier to an incorrect classification or simply suppress the correct classification. We also discuss the robustness of these patterns.

Hiding platforms in plain sight requires camouflage schemes that blend well with the environment. Such a camouflage scheme needs to cater for different geographical locations, seasons, and times of day. Inspired from nature’s biology, this paper presents a new algorithm, called Visible Signatures AI-generator (VSAI), for generating camouflage patterns iteratively to reduce visible signatures of objects. The proposed algorithm accepts a set of images from any dynamically changing environment. It then generates a customized set of camouflage patterns with colors and textures that are optimized for the environment. We present a novel Generative Adversarial Network (GAN), in which a generator with meta-parameters is iteratively trained to produce camouflage patterns. Simultaneously, a discriminator is trained to differentiate images with or without the embedded camouflage patterns. Unlike the existing methods, the meta-parameters used by our generator are intuitive, explainable, and extendable by the end-users. The experimental results show that the camouflage patterns designed by VSAI are consistent in color, texture, and semantic contents. Furthermore, VSAI produces improved outputs compared to several optical camouflage generation methods, including the Netherland Fractal Patterns, CamoGAN and CamoGen. The full end-to-end pattern generation process can operate at a speed of 1.21 second per pattern. Evaluated on the benchmark dataset Cityscapes, the YOLOv8 detector shows a significantly reduced target detection performance when our camouflage patterns are applied, yielding an mAP@0.5 detection score of 7.2% and an mAP@0.5:0.95 detection score of 3.2%. Compared to CamoGAN, our camouflage generation method leads to an average reduction of 4.0% in the mAP@0.5:0.95 detection score.

Thermal and visible cameras can be characterized by their Point Spread Function (PSF), which captures the aberrations induced by the image formation process, which includes effects due to diffraction or motion. Various techniques for estimating the PSF based on a simple image of a target object that consists of a random pattern were shown to be effective. Here, we describe a computational pipeline for estimating parametric Gaussian PSFs characterized by their width, height, and orientation, based on binary random pattern targets that are suitable for thermal imaging and easy to manufacture. Specifically, we consider the influence of deviating from a strict random pattern so the targets can be manufactured with common cutting or 3D printing devices. We evaluate the estimation accuracy based on simulated patterns with varying dot, pitch, and target sizes for different values of the point spread function parameters. Finally, we show experimental examples of acquired on manufactured devices. Our results indicate that the proposed random pattern targets offer a simple and affordable approach to estimating local PSFs.

In contrast to full simulation, ‘hybrid’ simulation, in which a virtual target is combined with a real recording of a scene, has been shown to create realistic imagery of targets with sufficient fidelity to assess the visual signature of a target. This method allows one to evaluate different camouflage designs and targets in a variety of backgrounds. Ultimately, the goal is to use a dataset of recorded imagery taken under various (weather) conditions as benchmark for signature analysis for any platform and camouflage pattern. To achieve this, the set of required calibration elements, such as MacBeth Color Checker and probe spheres, should be limited. In this study the extent to which a target coated with a given paint can be predicted from a limited set of painted probe spheres, e.g. differing in colour and/or gloss, is analyzed.

Training object detection algorithms to operate in complex geo-environments remains a significant challenge, necessitating large and diverse datasets (i.e., unique backgrounds and conditions) that are not always readily available. Physically generating requisite data can also be both cost and time prohibitive depending on the object(s) and area(s) of interest, especially in the case of multi-spectral and hyper-spectral imagery. Thus, there is increasing interest in the use of synthetic data to supplement existing physical datasets. To this end, the US Army Engineer Research and Development Center (ERDC) continues to develop a computational test-bed with a tool suite called the VESPA or, the Virtual Environmental Simulation for Physics-based Analysis, to support synthetic multi-spectral and hyper-spectral EO/IR imagery generation. The VESPA consists of integrated (1) scene generation tools, (2) multi-fidelity models for simulating heat and mass transfer and atmospheric energy propagation in geo-environments and climates worldwide that are optimized for high performance computing (3) data interrogation utilities, and (4) component-level sensor models capable of producing AI/ML ready near- and far-field imagery that is comparable to that produced by real sensors. This study presents an overview of the VESPA, new advances/capabilities, and results from a recent detailed validation and verification study.

Infrared imaging sensors can nowadays be regarded as a viable alternative to radar guidance that is stealthier and more capable of naval target classification, decoy discrimination and aimpoint selection. In view of this, the design of naval platforms, their sensors, weapon systems and counter measure deployment strategies need to be adapted accordingly. For this, tooling capable of simulating engagements by IR guided threats is essential. This paper presents a recently developed physics-based GPU-accelerated model chain that allows to generate realistic and radiometrically correct image sequences representative of those seen by an IR threat having varying levels of intelligence that is approaching a naval vessel. A description of the scientific and computational aspects of the model and modules will be provided along with examples of modelling output.

Space-Domain Awareness (SDA) via remote thermal imaging, where thermal-waveband EO/IR sensors are employed to observe orbiting satellites, has benefits over conventional visible/short-wave imaging. For example, LWIR sensors provide capability for both daytime and nighttime imaging, since temperature emissions and reflections are the basis of such observation (as opposed to optical sensors which rely on reflected light). To understand the capability that thermally dominant wavebands such as LWIR and MWIR can play in SDA, a robust simulation capability must be developed to predict signatures across the relevant spectrum. The computational complexity required for radiative transfer simulation is typically greater for satellite-focused thermal modeling in comparison to shortwave, reflected light-dominant wavebands. In this work, we employ MuSES to demonstrate the prediction of both internal and external temperature distributions for 3D satellite models. MuSES uses dynamic orbital boundary conditions to simulate transient solar loading, thermal radiation from Earth and to space, as well as radiative and conductive heat transfer from internal components such as electronics. Additionally, the coupled thermal/electrical multi-physics solvers in MuSES can incorporate realistic solar panel efficiency and battery cell charge/discharge cycling. Surfaces are attributed with spectral optical surface properties across the waveband(s) of interest to generate radiance maps via BRDF-based ray tracing of the predicted 3D temperature distributions. This allows radiometric signal levels of both the target and background, and subsequently contrast metrics of interest, to be generated with sensor simulations of space-based imaging platforms. Signature prediction is the primary output of this process, and in this study, we use our described methodology to demonstrate the inclusion of solar panel efficiency, battery charging/discharging and internal heat sources impact surface temperature distributions and infrared signatures during observations of satellites in LEO and GEO. A sensitivity study is performed to determine the significance that several satellite design choices can have on resultant signatures.

New Space Synthetic Aperture Radar (SAR) satellites yield an increasing potential for security and defense purposes, however, as of yet the lack of insight regarding sensor specific signature properties impedes this task. SAR simulation can provide a valuable insight into these signature properties. A selection of objects relevant for reconnaissance are explored regarding their signature characteristics in SAR images of New Space satellites, using the example of the Capella Space constellation. The focus is on temporary tank emplacements, their occupancy status and visibility. With the aid of Open-Source information, 3d models of temporary tank emplacements were generated for different states of occupancy. Using the CohRaS R SAR simulator of Fraunhofer IOSB these models were then simulated with sensor and acquisition parameters matching those of Capella Space. System specific properties, such as the variation of system noise levels or the limits of acquisition geometry are taken into account. Finally, a comparison between simulated signatures and signatures of real SAR imagery, as well as a discussion about the qualitative properties of these signatures are provided.

Hyperspectral atmospheric radiative transfer model (HARTM) is an essential component for image calibration and explanation in remote sensing applications. A HARTM describes the interaction between electromagnetic radiation and the earth’s atmosphere, which affects the quality and radiative accuracy of the acquired data. The performance evaluation of the HARTM is crucial to ensure the reliability of the retrieved information from calibrated images. By comparing the simulated results with measured or other validated reference data, the accuracy of HARTM can be assessed. Currently used similarity metrics, such as the Euclidean distance (ED) and the spectral angle metric (SAM), are relatively one-sided and single-valued overall assessment in evaluating hyperspectral model. The IEEE standard 1597.1 proposed feature selective validation (FSV) method as the key mathematical tool which has been widely applied in electromagnetic model verification, validation and accreditation (VV&A). However, to the best of our knowledge, the applications of FSV method in evaluating hyperspectral similarity has not yet been proposed up to this point. This paper concentrates on developing a technique for HARTM evaluation by means of FSV. Specifically, a multi-resolution components fused evaluation is proposed to obtain a more flexible comparison between the model and the measured data. As an example, the proposed approach is applied to validate the BeiHang University-Atmospheric Transfer Model (BHU-ATM) using the Michelson Interferometer for Passive Atmospheric Sounding (MIPAS) data. Results of multi-resolution components fused evaluation show good consistency with the results of directly evaluating the original data.

The proliferation of Unmanned Aerial Systems (UAS), both commercial-off-the-shelf and homebuilt, is increasingly posing a challenge to the authorities. Law enforcement agencies are confronted with the new task of policing the lower airspace, requiring them to be equipped with effective and appropriate counter-UAS systems for detecting, tracking, identifying (DTI) the offending drone. A number of technologies exists on the market today, using radar, EO/IR, acoustic and RF sensors. Evaluating the performance of these systems is not a trivial task, as different nations and entities often have different approaches. The COURAGEOUS project seeks to develop evaluation methodologies for counter-UAS systems, leading to a CENELEC pre-standard. We organized one of three validation trials for this project, implementing a maritime border protection scenario. We invited manufacturers to test their systems against a range of UAS, comparing their DTI data to the flightpath ground truth. These evaluations were not ranked to determine the best technology, but rather used to evaluate the usefulness of the evaluation methods developed by the project, allowing us to progress towards developing an EU-wide standard for evaluating DTI technologies.

The complex permittivity of adobe is measured using a coaxial probe system verses frequency (1 GHz to 4 GHz) and moisture content (0% to 6%). Measurements are performed using adobe samples collected from abode bricks. The variation of the adobe complex permittivity verses frequency is measured at discrete levels of moisture content using small adobe samples exposed to controlled levels of constant humidity in an environmental chamber. The typical moisture content profile verses depth for an adobe brick is also determined. It is shown that notable changes in material properties verses depth in the adobe wall results from moisture content variation in the adobe brick. Using the characterization of the adobe material, the application of Through-the-Wall Radar Imaging (TWRI) is considered for adobe walls. Matched illumination waveforms are derived, and the effects of optimal transmission waveforms are presented to illustrate the necessity of accurate material characterization for enhancement of TWRI applications. The results presented include simulation of an object located behind an adobe wall as well as experimental measurements taken in an anechoic chamber of an object located behind a wall section composed of adobe bricks. It is shown that enhanced TWRI performance may be obtained when utilizing knowledge of a material’s dielectric properties verses frequency and moisture level by reducing the amount of two-way attenuation of a radar waveform through waveform shaping techniques such as matched illumination waveform design.