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

With increasing threats to satellites and signal denial methods becoming cheap and effective, GPS failure is a reality and critical risk for navigation, localization, and targeting applications. Inspired by nature, the SkyPASS polarimeter developed by Polaris Sensor Technologies exploits the atmospheric polarization pattern to find highly accurate heading in situations when a typical sun/star sensor would fail to operate. It provides improved availability under cloud cover, under canopy, in urban environments, in civil and nautical twilight, and during sunrise and sunset. Unpolarized sunlight (or moonlight) becomes partially polarized when scattered by atmospheric molecules. Rayleigh scattering creates a polarization pattern, or map, that is unique, depending upon the date, time, and the position of the observer. This natural phenomenon is the scientific basis for SkyPASS operation, and it can be predicted to first order using Rayleigh scattering theory. This paper provides an overview of the SkyPASS polarimeter design, the method to calculate heading from sky polarization information, and the performance of the polarimeter in different environments. The third generation of SkyPASS processes data in real-time and has small enough SWaP to fit almost any platform.

Over the past decade, a large body of work has demonstrated improved designs over the conventional 2×2 modulation scheme (based upon polarizer orientation angles of {0, 90, 45, 135}o) used in the manufacture of DoFP polarimeters. These designs show better usage of bandwidth and reduction of crosstalk across the various linear Stokes vector channels in the frequency domain. While much focus has been on the development of optimal modulation schemes for these devices, little attention has been given to the development of corresponding demosaicing strategies for these modulators. In this work, we adapt a recent demosaicing strategy based upon a conditional generative adversarial network (cGAN) developed for conventional 2×2 DoFP sensors to accommodate alternative modulation schemes. We collect full-resolution polarized intensity data at non-conventional polarizer angles using a division-of-time (DoT) visible imaging polarimeter that we use to simulate DoFP data from alternative modulators for training and testing purposes. We then assess performance across these alternative modulation strategies and compare results against the conventional modulation scheme.

We recently presented a deep learning approach to demosaic division of focal plane (DoFP) imaging polarimeter data based upon a conditional generative adversarial network (cGAN). The approach was developed and demonstrated using visible DoFP polarimeter data and showed a notable ability to reduce false edge artifacts, aliasing, and temporal noise. Here we retrain and apply this algorithm to emissive-band polarimetric data acquired with a LWIR DoFP imaging polarimeter to investigate performance. We then adapt the baseline cGAN architecture to perform simultaneous demosaicing and resolution enhancement of LWIR DoFP data. We collect full-resolution polarized intensity data using a division-of-time (DoT) LWIR imaging polarimeter that we use to simulate decimated DoFP data for training and testing purposes. We then apply the algorithm to data obtained from simulated LWIR DoFP polarimeter data and assess performance.

Division of Focal plane imagers have recently been developed for polarimetric imaging. Those sensors use a grid composed of four different pixels with four different polarizers engraved on them. Four of these different pixels form a superpixel which enables the estimation of the linear Stokes vector with a single acquisition. Those sensors are particularly sensitive to the spatial variations of the scene. Therefore, if such variations are non-negligible compared with the measurement noise, the estimation of the state of polarization is corrupted. We propose a method to map the superpixels in which the estimation can be trusted.

We have developed a polarization sensitive imaging device that provides both synchronous (ie frame based) and asynchronous (ie event based) visual information. The sensor has a dynamic range over 120 dB and has sub-millisecond latency. The event based polarization information is processed with a deep neural network in order to reconstruct angel and degree of polarization data beyond the maximum Nyquist frame rate .

MODTRAN7, a polarimetric extension of the MODTRAN6 atmospheric radiative transfer model, is being developed. The vectorized MODTRAN7 will provide band model (BM), correlated-k (Ck), and line-by-line options for computing Stokes vectors. The radiative transfer problem is being solved for Isotropic and Symmetric Media (ISM) using the basic phenomenology described in the classic text by Mishchenko, Travis and Lacis, “Multiple Scattering of Light by Particles”¹. VDISORT, a vectorized version of the DISORT scalar model currently in MODTRAN6, will compute the Stokes vectors for 1-D atmospheres. The MODTRAN method for extracting spherical refractive path contributions from the plane parallel scattering models will be adapted for the polarimetric model. The upgrade is to include new polarimetric optical properties for both the existing aerosol and cloud models within MODTRAN and for recently developed cirrus cloud and dust particulate data. A new algorithm has been developed that enables Generalized Spherical Function (GSF) expansion coefficients to be accurately computed to very high order. MODTRAN has already been restructured to generate Stokes vector data for single scatter solar/lunar applications and validated against the NASA Goddard Space Flight Center model, 6SV.

While Spectralon panels are largely assumed to be ideal Lambertian surfaces, their actual polarized reflective responses deviate from the ideal by at least a small amount at illumination and viewing angles off surface normal. The Mueller matrix response of four different panels between 10% and 99% reflectance were measured and the radiometric response from two distinct monostatic or nearmonostatic polarimeter systems are compared, one at Montana State University and one at the Air Force Research Lab. The deviations from an assumed ideal Lambertian surface are reported.

It is well known that thermal cameras augmented with polarization sensing are often better at helping a user find man-made objects in clutter than thermal only cameras. Thermal imaging polarimetry is an emerging technology and sensors are now commercially available. As this technology becomes more widely used, early adopters often want to compare performance of their thermal/polarization cameras operating in thermal and thermal/polarimetric mode. As such, metrics are needed to make a fair and rigorous comparison. In this paper, we introduce three visibility metrics that compare the average intensity, internal contrast, and spatial frequencies of the target to that of local background. The visibility of several targets is compared in various backgrounds to illustrate the utility of the metrics. To help validate the metrics, target visibility scoring using the metrics is compared to human visibility scoring. Finally, the visibility metrics are used for clutter analysis where the target visibility is ranked according to its similarity to scene clutter objects.

We discovered a theoretical link between the reflectivity of a solid surface and the resulting polarization, and we want to exploit this finding to advance the state-of-the-art of polarimetric remote sensing. At a special geometry, a spectral plot of polarization versus reflectivity collapses into a 1-dimensional degenerate curve (termed the U-curve) that applies to all materials, both conductors and dielectrics. The U-curve shows an inverse relationship between polarization and reflectivity and provides a new theoretical underpinning to explain why dark objects are more polarizing than brighter ones.1-7 We argue that the U-curve represents the Polarization Relative to the Maximum Attainable (PReMA) for a surface of given reflectivity. We claim that standard polarization metrics (e.g., DoLP) are biased because most surfaces cannot polarize reflected light at the 100% level, and therefore, by using PReMA as a constraint for normalization, we can create a new polarization metric that should be superior to old ones. Because the U-curve applies to perfectly smooth dielectric materials, in theory, polarimetric measurements at this special geometry could be used to optically quantify the surface roughness of a material. Also, we propose that the PReMA method could provide the basis for a new, bi-static radar observation scenario to provide more quantitative information about the earth’s surface.

An object’s polarimetric bidirectional reflection distribution function (pBRDF) is fully parameterized by the 16 degrees of freedom of a Mueller matrix (MM) at each scattering geometry. A common pBRDF approximation to reduce the degrees of freedom is as a weighted sum of a Fresnel reflection term and an ideal depolarizer term. The weights on these terms represent fractional specular and diffuse reflection and are typically fit independently. Any MM for which the smallest three eigenvalues of the Cloude MM decomposition are identical,¹ can be rewritten as a convex sum of a dominant non-depolarizing MM and an ideal depolarizer.^{2, 3} Therefore, the fractional contribution of each term in this pBRDF model is a single depolarization parameter which corresponds to the largest eigenvalue.² The reduced degrees of freedom for pBRDFs described by this single depolarization parameter create an opportunity to utilize partial polarimetry. The primary contribution of this work is a linear estimator for a MM’s dominant eigenvalue which requires fewer measurements than a full MM reconstruction. Despite reducing the number of simulated measurements by a factor of 10, partial-polarimetry and full Mueller polarimetry eigenvalue estimates are comparable. Root-mean-squared error (RMSE) averaged over acquisition geometry for eigenvalues of a white and a gray balance card were 0.027 and 0.025 respectively for 4 polarimetric measurements, and 0.019 and 0.032 respectively for 40 polarimetric measurements. MM extrapolations from measurements with a commercial off-the-shelf linear Stokes camera are performed at 25 acquisition geometries on an ensemble of LEGO bricks treated to have varying surface roughness. Averaged over the acquisition geometries, the partial-polarimetry extrapolated MMs achieve a 7.3% minimum and 15.1% maximum flux discrepancy from full-polarimetry reconstructed MMs over the varying surface textures. This work demonstrates the first approach, known to the authors, for extrapolating depolarizing MMs.

Polarization-sensitive microscopy has been a hot topic in biology over the last decades since many biological samples, including cancer cells and tendons, are intrinsically birefringent. Among its different applications, polarization-sensitive microscopy has enabled the detection and classification of diseases in biological samples by enabling the analysis of their polarimetric properties. In this contribution, we present an apparatus and method to estimate the Mueller matrix of microscopic samples using intensity-based images recorded from a bright-field microscope. We have validated the proposed technique using standard linear polarizers. The hallmark characteristics of our technique are its simplicity and efficiency, being suitable for analyzing biological and other birefringent samples.

Physics-based rendering (PBR) engines attempt to generate photorealistic images by mimicking light-matter interaction in a physically plausible way. PBR has become the standard rendering method in the fields of animation, gaming, and computer graphics research. More recently, PBR engines have included the ability to track the full polarization state of light. An area of interest for polarization-aware PBR engines is validating the accuracy of polarized bi-direction reflection distribution functions (pBRDF). pBRDFs are polarized material models described by a geometry-, texture-, and albedo-dependent Mueller matrix. For renderings, methods to analyze the pBRDF are limited. This work presents a pBRDF analysis method that simulates a Mueller matrix imaging polarimeter using a polarization-aware PBR engine. Simulated reconstructed Mueller matrix images are qualitatively compared to measurements from a Mueller matrix imaging polarimeter.

Polarization imaging has been used extensively in applications related to atmospheric monitoring, remote sensing, and quality control. However, it has been used less extensively in agricultural applications, where color sensing - either red, green, and blue (RGB) imaging, multispectral, and/or hyperspectral cameras are more common. In this paper, we discuss our preliminary results related to the use of polarization imaging to quantify defoliation in peanut plants in response to leaf spot disease. A key metric for breeding resistant peanut varieties involves identifying the point at which defoliation occurs. Since defoliation is a geometrical property of the plant canopy, we investigated whether polarization imaging can provide a better-automated score when compared to conventional visual scoring. Initial results are presented, as well as a discussion of our drone-based platform and our experimental trials conducted during the 2021 North Carolina growing season.

We introduce two new tools to the application of polarimetry to space domain awareness (SDA), the LoVIS spectropolarimeter on the 3.6 m AEOS telescope and deep convolutional neural networks (CNNs). Using a dataset of 20,000 simulated satellite observations, we train a CNN to map distance-invariant spectropolarimetric data to object identity. We report the classification accuracy of this simulation for a 9-class satellite problem, comparing results against low-resolution spectra for which prior success has been demonstrated as well as solar phase angle and satellite apparent magnitude. These initial experiments show potential for improved discrimination against nearly identical satellites on the basis of added polarimetric data.

The degree of linear polarization of above-water radiance contains information on the attenuation-to-absorption ratio in the surface layer of the ocean. With additional inversion algorithms to retrieve size distribution, backscattering ratio and bulk refractive index of marine particle populations, polarization is thus an important component in the retrieval of optical and microphysical properties in the ocean. Accordingly, in preparation of the upcoming PACE mission, we used an artificial dataset of water IOPs and atmospheric properties in conjunction with radiative transfer simulations to explore the potential of a neural network approach for systematic conversion of top-of-atmosphere polarization values first into above-water polarization, then finally into an estimation of local optical and microphysical properties. Additionally, we used polarization images acquired in situ during a cruise in the Gulf of Mexico to study wave slope statistics as derived from a modified polarimetric slope sensing technique and found them to compare well with the classic wave slope variances of Cox and Munk.

We propose polarimetric calibration of nano-satellites by pointing them towards solar panel farms or mirrors at solar thermal power plants. We show through simulations that both can provide significant polarization sources. Around a solar tower, this is obtained by polarized skylight reflected from mirrors. Photovoltaic solar panels, on the other hand, yield a strong polarized signal by reflecting direct sunlight around the Brewster angle. The signal is affected by aerosols. The aerosol uncertainty affects calibration tasks. Based on these findings, we simulate spaceborne polarimetric camera calibration.

Interactance spectroscopy is a relative scattering technique that is useful for measuring the internal spectral absorption characteristics of agricultural produce. However, deploying it in high-speed scanners is challenging due to the limited radiometric output of broadband white-light sources. Further complexities relate to the need to mask the incident first-surface reflected light from the detector, so more deeply-scattered interacting photons can be measured. In this paper, we describe a multispectral laser-based line-projection configuration that enables interactance measurements without requiring an obscuration. Preliminary low-throughput and high spectral resolution measurements were taken to inform the choice of laser wavelengths. Polarization cameras are used during scanning to aid in identifying regions of maximum scattering to reduce the impact of first-surface reflections. We discuss the system’s optical design and demonstrate preliminary results from the scanner.

An InfraRed Channeled Spectro-Polarimeter (IRCSP) was demonstrated in the near space environment as a piggyback out of NASA Columbia Scientific Ballooning Facility. The compact IRCSP is sensitive to linearly polarized long-wave infrared (LWIR) light between 7-12 microns and targets cloud micro-physical properties. Post landing the instrument was retrieved with no damage to the optical payload and collected over 150 minutes of flight data at altitudes above 30 km. The results collected both demonstrate the operation of uncooled microbolometers in the low pressure environment and are the first know high-altitude observations of a polarized signal from cloud tops in the LWIR. During deployment, the IRCSP reported brightness temperatures between 250-285K with uncertainty of < 1:5K. In addition, statistically significant polarization modulation with degrees of linear polarization (DoLP) between 1 – 20% and preferential angle of linear polarization (AoLP) trends were detected. These results support the hypothesis that the LWIR polarimetry has the potential to add new sensitivity to existing remote sensing platforms.

The Digital Imaging and Remote Sensing Image Generation (DIRSIG) model is a physics based image and data generation tool developed at the Rochester Institute of Technology. While DIRSIG5 includes a plug-in based architecture and a path-tracing approach to determine the aperture reaching radiance, it does not support polarized radiometric propagation at this time. DIRSIG4 does include this polarized radiometric propagation via Stokes vectors, Mueller matrices, and parameterized polarized bi-directional reflectance distribution functions (pBRDFs). This paper includes an overview of the pBRDF models available to the user, sources of polarimetric reflectance data, and how DIRSIG4 generates a Stokes vector radiance output image. Example imagery of custom targets and full scale scenes will be included to aid users in setting up their own simulations for testing algorithms that require polarized imagery.

Urban material discrimination and other infrastructure assessment can be difficult to perform remotely. Polarimetry has been shown to aid in discriminating between material classes, but only recently has been studied in the context of discrimination within classes of construction materials. We present new results focused on discriminating between nine concrete materials. The materials were illuminated at varied elevation angles at visible and near-infrared wavelengths and imaged from nadir. We show that analyses of multispectral polarimetric images can provide useful discrimination where multispectral images alone cannot.