A computationally efficient radiative transport model is presented that predicts a camera measurement and accounts for the light reflected and blocked by an object in a scattering medium. The model is in good agreement with experimental data acquired at the Sandia National Laboratory Fog Chamber Facility (SNLFC). The model is applicable in computational imaging to detect, localize, and image objects hidden in scattering media. Here, a statistical approach was implemented to study object detection limits in fog.
Performing terrain classification with data from heterogeneous imaging modalities is a very challenging problem. The challenge is further compounded by very high spatial resolution. (In this paper we consider very high spatial resolution to be much less than a meter.) At very high resolution many additional complications arise, such as geometric differences in imaging modalities and heightened pixel-by-pixel variability due to inhomogeneity within terrain classes. In this paper we consider the fusion of very high resolution hyperspectral imaging (HSI) and polarimetric synthetic aperture radar (PolSAR) data. We introduce a framework that utilizes the probabilistic feature fusion (PFF) one-class classifier for data fusion and demonstrate the effect of making pixelwise, superpixel, and pixelwise voting (within a superpixel) terrain classification decisions. We show that fusing imaging modality data sets, combined with pixelwise voting within the spatial extent of superpixels, gives a robust terrain classification framework that gives a good balance between quantitative and qualitative results.
Random scattering and absorption of light by tiny particles in aerosols, like fog, reduce situational awareness and cause unacceptable down-time for critical systems or operations. Computationally efficient light transport models are desired for computational imaging to improve remote sensing capabilities in degraded optical environments. To this end, we have developed a model based on a weak angular dependence approximation to the Boltzmann or radiative transfer equation that appears to be applicable in both the moderate and highly scattering regimes, thereby covering the applicability domain of both the small angle and diffusion approximations. An analytic solution was derived and validated using experimental data acquired at the Sandia National Laboratory Fog Chamber facility. The evolution of the fog particle density and size distribution were measured and used to determine macroscopic absorption and scattering properties using Mie theory. A three-band (0.532, 1.55, and 9.68 μm) transmissometer with lock-in amplifiers enabled changes in fog density of over an order of magnitude to be measured due to the increased transmission at higher wavelengths, covering both the moderate and highly scattering regimes. The meteorological optical range parameter is shown to be about 0.6 times the transport mean free path length, suggesting an improved physical interpretation of this parameter.
Deciding on an imaging modality for terrain classification can be a challenging problem. For some terrain classes a given sensing modality may discriminate well, but may not have the same performance on other classes that a different sensor may be able to easily separate. The most effective terrain classification will utilize the abilities of multiple sensing modalities. The challenge of utilizing multiple sensing modalities is then determining how to combine the information in a meaningful and useful way. In this paper, we introduce a framework for effectively combining data from optical and polarimetric synthetic aperture radar sensing modalities. We demonstrate the fusion framework for two vegetation classes and two ground classes and show that fusing data from both imaging modalities has the potential to improve terrain classification from either modality, alone.
This communication reports progress towards the development of computational sensing and imaging methods that utilize highly scattered light to extract information at greater depths in degraded visual environments like fog for improved situational awareness. As light propagates through fog, information is lost due to random scattering and absorption by micrometer sized water droplets. Computational diffuse optical imaging shows promise for interpreting the detected scattered light, enabling greater depth penetration than current methods. Developing this capability requires verification and validation of diffusion models of light propagation in fog. We report models that were developed and compared to experimental data captured at the Sandia National Laboratory Fog Chamber facility. The diffusion approximation to the radiative transfer equation was found to predict light propagation in fog under the appropriate conditions.
There are several factors that should be considered for robust terrain classification. We address the issue of high pixel-wise variability within terrain classes from remote sensing modalities, when the spatial resolution is less than one meter. Our proposed method segments an image into superpixels, makes terrain classification decisions on the pixels within each superpixel using the probabilistic feature fusion (PFF) classifier, then makes a superpixel-level terrain classification decision by the majority vote of the pixels within the superpixel. We show that this method leads to improved terrain classification decisions. We demonstrate our method on optical, hyperspectral, and polarimetric synthetic aperture radar data.
The detection, location, and identification of suspected underground nuclear explosions (UNEs) are global security priorities that rely on integrated analysis of multiple data modalities for uncertainty reduction in event analysis. Vegetation disturbances may provide complementary signatures that can confirm or build on the observables produced by prompt sensing techniques such as seismic or radionuclide monitoring networks. For instance, the emergence of non-native species in an area may be indicative of anthropogenic activity or changes in vegetation health may reflect changes in the site conditions resulting from an underground explosion. Previously, we collected high spatial resolution (10 cm) hyperspectral data from an unmanned aerial system at a legacy underground nuclear explosion test site and its surrounds. These data consist of visible and near-infrared wavebands over 4.3 km2 of high desert terrain along with high spatial resolution (2.5 cm) RGB context imagery. In this work, we employ various spectral detection and classification algorithms to identify and map vegetation species in an area of interest containing the legacy test site. We employed a frequentist framework for fusing multiple spectral detections across various reference spectra captured at different times and sampled from multiple locations. The spatial distribution of vegetation species is compared to the location of the underground nuclear explosion. We find a difference in species abundance within a 130 m radius of the center of the test site.
We present simulation results quantitatively showing that circularly polarized light persists in transmission through several real-world and model fog environments better than linearly polarized light over broad wavelength ranges from the visible through the infrared. We present results for polydisperse particle distributions from realistic and measured fog environments, comparing the polarization persistence of linear and circular polarization. Using a polarization-tracking Monte Carlo program, we simulate polarized light propagation through four MODTRAN fog models (moderate and heavy radiation fog and moderate and heavy advection fog) and four real-world measured fog particle distributions (Garland measured radiation and advection fogs, Kunkel measured advection fog, and Sandia National Laboratories’ Fog Facility’s fog). Simulations were performed for each fog environment with wavelengths ranging from 0.4 to 12 μm for increasing optical thicknesses of 5, 10, and 15 (increasing fog density or sensing range). Circular polarization persists superiorly for all optical wavelength bands from the visible to the long-wave infrared in nearly all fog types for all optical thicknesses. Throughout our analysis, we show that if even a small percentage of a fog’s particle size distribution is made up of large particles, those particles dominate the scattering process. In nearly all real-world fog situations, these large particles and their dominant scattering characteristics are present. Larger particles are predominantly forward-scattering and contribute to circular polarization’s persistence superiority over broad wavelength ranges and optical thicknesses/ range. Circularly polarized light can transmit over 30% more signal in its intended state compared to linearly polarized light through real-world fog environments. This work broadens the understanding of how circular polarization persists through natural fog particle distributions with natural variations in mode particle radius and single or bimodal characteristics.
Scattering environment conditions, such as fog, pose a challenge for many detection and surveillance active sensing operations in both ground and air platforms. For example, current autonomous vehicles rely on a range of optical sensors that are affected by degraded visual environments. Real-world fog conditions can vary widely depending on the location and environmental conditions during its creation. In our previous work we have shown benefits for increasing signal and range through scattering environments such as fog utilizing polarized light, specifically circular polarization. In this work we investigate the effect of changing fog particle sizes and distributions on polarization persistence for both circularly and linearly polarized light via simulation. We present polarization tracking Monte Carlo results for a range of realistic monodisperse particle sizes as well as varying particle size distributions as a model of scattering environments. We systematically vary the monodisperse particle size, mean particle size of a distribution, particle size distribution width, and number of distribution lobes (bi-modal), as they affect polarized light transmission through a scattering environment. We show that circular polarization signal persists better than linear polarization signal for most variations of the particle distribution parameters.
Degraded visual environments are a serious concern for modern sensing and surveillance systems. Fog is of interest due to the frequency of its formation along our coastlines disrupting border security and surveillance. Fog presents hurdles in intelligence and reconnaissance by preventing data collection with optical systems for extended periods. We will present recent results from our work in operating optical systems in our controlled fog experimental chamber. This facility is a 180-foot-long, 10-foot-wide, and 10-foot-Tall structure that has over 60 spray nozzles to achieve uniform aerosol coverage with various particle size, distributions, and densities. We will discuss the physical formation of fog in nature and how our generated fog compares. In addition, we will discuss fog distributions and characterization techniques. We will investigate the biases of different methods and discuss the different techniques that are appropriate for realistic environments. Finally, we will compare the data obtained from our characterization studies against accepted models (e.g., MODTRAN) and validate the usage of this unique capability as a controlled experimental realization of natural fog formations. By proving the capability, we will enable the testing and validation of future fog penetrating optical systems and providing a platform for performing optical propagation experimentation in a known, stable, and controlled environment.
We present simulation and experimental results showing circular polarization is more tolerant of optical collection geometry (field of view and collection area) variations than linear polarization for forward-scattering environments. Circular polarization also persists superiorly in the forward-scattering environment compared to linear polarization by maintaining its degree of polarization better through increasing optical thicknesses. In contrast, both linear and circular polarizations are susceptible to collection geometry variations for isotropic-scattering (Rayleigh regime) environments, and linear polarization maintains a small advantage in polarization persistence. Simulations and measurements are presented for laboratory-based environments of polystyrene microspheres in water. Particle diameters were 0.0824 μm (for isotropic-scattering) and 1.925 μm (for forward-scattering) with an illumination wavelength of 543.5 nm.
This work supports Airborne Intelligence, Surveillance, and Reconnaissance (ISR) for tactical situational awareness in challenging environments with modified imaging LIDAR (light detection and ranging). LIDAR produces an irradiance-based scene with high, three-dimensional, spatial resolution; differentiating reflecting surfaces and surface textures not just for target detection, but also target recognition. LIDAR is generally prevented from working through all weather; as the traditional source wavelengths are scattered and/or absorbed by fog, clouds, and dust known as degraded visual environments (DVEs). This work identifies and quantifies improved optical wavelength regimes and polarization strategies that should open this otherwise denied operating window for LIDAR. We demonstrate modified imaging LIDAR's utility and ability to produce images in environments that have been challenging for traditional LIDAR (fog, dust) systems. We utilize a state-of-the-art Geiger mode avalanche photodiode (GMAPD) 32X32 detecting array for imaging with an integrated fast timing circuit ROIC per imaging detector pixel. This GMAPD is equivalent to 1024 radar receivers and produces a 3-D point cloud scene for each %3Clnsec laser interrogation pulse. Through this work, we have demonstrated Sandia's first imaging LIDAR system. We developed significant software to scan the GMAPD array, control data acquisition, and stitch an arbitrary number of images for greater lateral spatial resolution. We designed and built an optical system to evaluate performance, in the laboratory and in field tests, and relate to the performance at tactically important distances. Our use of this system is to study and characterize techniques that extend the range of useful imaging through fog for applications in a maritime environment. We quantify, with simulations, broad wavelength regimes and illuminating polarizations that have benefit in propagating through fog. In particular, we identify circular polarization over broad wavebands as a superior persisting signal technique in extending the useful range of active LIDAR in forward-scattering environments. Finally, we characterize the performance of our LIDAR system and confirm our theoretical predictions in the field and directly evaluate this system as a function of polarization in Sandia's medium-range Fog Tunnel Facility. This technology has potential impacts in improved ISR in maritime environments, but with further development could be revolutionary in tactical ISR by providing real-time three-dimensional full motion video (3D FMV) in DVEs.
We present experimental and simulation results for a laboratory-based forward-scattering environment, where 1 μm diameter polystyrene spheres are suspended in water to model the optical scattering properties of fog. Circular polarization maintains its degree of polarization better than linear polarization as the optical thickness of the scattering environment increases. Both simulation and experiment quantify circular polarization's superior persistence, compared to that of linear polarization, and show that it is much less affected by variations in the field of view and collection area of the optical system. Our experimental environment's lateral extent was physically finite, causing a significant difference between measured and simulated degree of polarization values for incident linearly polarized light, but not for circularly polarized light. Through simulation we demonstrate that circular polarization is less susceptible to the finite environmental extent as well as the collection optic's limiting configuration.
This work quantifies the polarization persistence and memory of circularly polarized light in forward-scattering and isotropic (Rayleigh regime) environments; and for the first time, details the evolution of both circularly and linearly polarized states through scattering environments. Circularly polarized light persists through a larger number of scattering events longer than linearly polarized light for all forward-scattering environments; but not for scattering in the Rayleigh regime. Circular polarization's increased persistence occurs for both forward and backscattered light. The simulated environments model polystyrene microspheres in water with particle diameters of 0.1 μm, 2.0 μm, and 3.0 μm. The evolution of the polarization states as they scatter throughout the various environments are illustrated on the Poincaré sphere after one, two, and ten scattering events.
We present simulation results that show circularly polarized light persists through scattering environments better than linearly polarized light. Specifically, we show persistence is enhanced through many scattering events in an environment with a size parameter representative of advection fog at infrared wavelengths. Utilizing polarization tracking Monte Carlo simulations we show a larger persistence benefit for circular polarization versus linear polarization for both forward and backscattered photons. We show the evolution of the incident polarization states after various scattering events which highlight the mechanism leading to circular polarization's superior persistence.
We find for infrared wavelengths that there are broad ranges of particle sizes and refractive indices that represent fog and rain, where circular polarization can persist to longer ranges than linear polarization. Using polarization tracking Monte Carlo simulations for varying particle size, wavelength, and refractive index, we show that, for specific scene parameters, circular polarization outperforms linear polarization in maintaining the illuminating polarization state for large optical depths. This enhancement with circular polarization can be exploited to improve range and target detection in obscurant environments that are important in many critical sensing applications. Initially, researchers employed polarizationdiscriminating schemes, often using linearly polarized active illumination, to further distinguish target signals from the background noise. More recently, researchers have investigated circular polarization as a means to separate signal from noise even more. Specifically, we quantify both linearly and circularly polarized active illumination and show here that circular polarization persists better than linear for radiation fog in the short-wave infrared, for advection fog in the short-wave and long-wave infrared, and large particle sizes of Sahara dust around the 4 μmwavelength. Conversely, we quantify where linear polarization persists better than circular polarization for some limited particle sizes of radiation fog in the long-wave infrared, small particle sizes of Sahara dust for wavelengths of 9-10.5 μm, and large particle sizes of Sahara dust through the 8-11 μm wavelength range in the long-wave infrared.
We find for infrared wavelengths there are broad ranges of particle sizes and refractive indices that represent fog and rain where the use of circular polarization can persist to longer ranges than linear polarization. Using polarization tracking Monte Carlo simulations for varying particle size, wavelength, and refractive index systematically, we show that for specific scene parameters circular polarization outperforms linear polarization in maintaining the intended polarization state for large optical depths. This enhancement in circular polarization can be exploited to improve range and target detection in obscurant environments that are important in many critical sensing applications. Specifically, circular polarization persists better than linear for radiation fog in the short-wave infrared, for advection fog in the short-wave infrared and the long-wave infrared, and large particle sizes of Sahara dust around the 4 micron wavelength.