Polarization based detection is often accomplished by using two separate components, reflectivity/emissivity and polarization, as detection algorithm inputs. These are Stokes vector components and are derived from elementary factors that represent energy collected with different polarizers. The elementary factors are added to produce the reflectivity/emissivity component and subtracted to produce the polarization component. Using the reflectivity/emissivity and polarization clearly addresses the advantage of using polarization as an added discriminant. However, depending on the detection algorithm, it may be better to use the elementary factors as input into a detection algorithm. A constant false alarm rate detection algorithm derived from a maximum likelihood is used as a foundation for judging target detection with these two different inputs. The results are presented for detecting man-made objects on natural backgrounds. The data cover two incident light sources: natural light, which is unpolarized and a linearly polarized laser. Detection using the elementary factors is shown to be consistent with detection using the Stokes vector components and is shown to decrease the false alarm rate.

We develop new techniques for polarization-encoded fingerprint images that correspond to Stokes vector imagery. We show how the polarization-encoded images can effectively be used to increase the discrimination capabilities of fingerprint systems for security applications in defense, law enforcement and civilian areas. Computer simulation supported by optical experiments is also presented to further support the validity of the proposed technique.

We examine the joint probability distribution of polarization information; specifically the joint distribution of the degree of polarization and the three Stokes parameters, as it relates to material properties in a highly scattering, highly depolarizing transmission geometry, making use of the Observable Polarization Sphere (OPS) for visualization of this distribution. We comment on the role of the source’s coherence in terms of the polarization properties of the resulting speckle field, and describe numerical and experimental results of second order Stokes vector element correlations and their applicability to material discrimination.

We have developed a scattering and emission polarization model. The scattering model is based on the vector Kirchhoff diffraction equation. For the emission polarization model, we use the Kirchhoff law for opaque materials, where the directional emissivity is assumed to be the same as Fresnel parallel and perpendicular transmittance. It is assumed that the emitted radiation from each facet has no relation with the radiation from neighboring facets. The roughness of the surface is treated statistically using the rms roughness height and the autocorrelation length. A Gaussian distribution is assumed for the roughness facet normal vectors. Shadowing by neighboring facets is also included in the model. We compute look-up tables for the scattering Mueller matrix and emission Stokes parameters for all incident and scattering (or emission) angles. The look-up tables are used for simulating the scattering and emission polarization signatures of objects. Polarization images of scattering, self-emission, and combination of scattering and emission are studied for Aluminum objects of various roughnesses and for various wavebands. The results show that the surface roughness is an important factor to determine the intensity and polarization. Our simulation results agree with polarization field data in that solar reflection has larger cancellation effect on MW IR polarization signatures than LW IR.

The discrimination of scene elements in polarimetric and in non-polarimetric images is governed by both environmental and instrumental factors. These factors consist of systematic elements, which are dealt with by means of appropriate calibration, and random errors. In the case of imaging polarimetry, the Stokes parameter images are calculated from images obtained with orthogonal orientations of the linear polarizer about the optic axis. For the stokes images to contain significant information, the orthogonal, registered image pair from which the Stokes images S₁ and S₂ are calculated must be significantly different. Misregistration of the orthrogonal input images also impacts the correlation of the resulting Stokes image to scene elements. The system MTF, sampling patter and geometry further impact the discrimination of features in the scene. These factors are discussed. The effects of systematic and random error sources on resolved target discriminability from clutter background is considered in depth. While the issue of spatially unresolved target detection is considered, it does not form a major component of this discussion. The intent of these considerations of the physics and phenomenology of imaging polarimetry is to progress towards the predictive modeling of target discriminability. This will aid in sensor design and mission parameter optimization.

Optical properties of vegetation materials are described. Spectral measurements of reflectance and Mueller matrices in the near infrared are given. The measurement method is discussed, and results are presented for a variety of plant material. Samples include both green leaves and bark. Measured results are compared to published results were possible.

Polarization measurements in the infrared region have earlier been done on cenosphere surfaces. These surfaces have been constructed so that they have almost completely depolarized reflected and emitted radiations. It was also possible to construct the surface so that a desired normal incidence reflectivity and emissivity was achieved. The measurements were performed by measuring emissivity as a function of the emissive angle, from 5° to 85°, and for different polarization angles. From these angle measurements the polarization parameters such as Degree of Linear Polarization (DoLP) and θ can been calculated. In this work, model calculations of scattering from a rough surface consisting of cenospheres are performed, using Monte Carlo calculations on a geometrical optical model. The model calculations have given good results and can explain both the depolarization and the possibility to determine the emissivity, from the cenosphere surface, by changing the amount of gold deposited on the surface.

The automated, or semi-automated analysis of scene elements in a clutter background is more complex in polarimetric imaging than in conventional imaging. This is largely due to the fact that misregistration of the orthogonal images used to calculate the Stokes parameter images introduces an artificial clutter. Further, there is little reported information on polarimetric image clutter. We present representative findings from an analysis of polarimetric image data, obtained over various backgrounds with various geometries, and examine the manner in which systematic and random variations impact feature discriminations.

The factors governing the extraction of useful information from polarimetric images depend upon the image acquisition and analytical methodologies being used, and upon systematic and environmental variations present during the acquisition process. The acquisition process generally occurs with foreknowledge of the analysis to be used. Broadly, interactive image analysis and automated image analysis are two different procedures: in each case, there are technical challenges. Imaging polarimetry is more complex than other imaging methodologies, and produces an increased dimensionality. However, there are several potential broad areas of interactive (manual) and automated remote sensing in which imaging polarimetry can provide useful additional information. A review is presented of the factors controlling feature discrimination, of metrics that are used, and of some proposed directions for future research.

The identification of terrestrial objects using hyperspectral measurements is confounded by the influence of the intervening atmosphere that can obscure spectral features through gaseous line absorption and reduce contrast as a result of scattering by aerosols. Although hyperspectral measurements provide an effective source for the estimation of the amount of water vapor the estimate of aerosol properties using spectral radiances is much more uncertain. This deficiency can be remedied by making polarimetric measurements which allow a more accurate and complete retrieval of aerosol properties over land than simple radiance measurements. As a proof of this concept we have made measurements of known ground targets with the HyperSpecTIR (HST), a flexible, airborne hyperspectral imager capable of on-the-fly programmability and the Research Scanning Polarimeter (RSP). Measurements were made both in the mountains near Santa Barbara and at lower altitude near Buellton. The collected data is analyzed, atmospherically corrected and compared to the known spectral reflectance of the ground targets. The capabilities and deficiencies of the measurements, analysis and atmospheric correction technique are discussed.

Polarized light plays an important role in the underwater environment. Light that is scattered within the water is partially polarized. Biological and artificial systems can exploit this phenomenon. We aim to utilize this phenomenon in a new generation of underwater imaging systems in order to partially compensate for the loss of color and visibility. In order to obtain quantitative measurement of radiance and polarization, the imaging system should have a linear radiometric response and low noise. In addition, the interface of the camera with the water should have a minimum effect on the polarization. In this paper, we describe a portable lightweight imaging system that addresses these conditions. We detail the design considerations and empirical verifications.

Current electro-optical based landmine detection techniques focus on exploiting phenomena across several wavebands. In particular, polarization signatures have and continue to be a focus of interest for this problem. Our research examined these signatures in the context of a real world environment; specifically, we examined the spectral polarization characteristics of landmines and soils within a complex radiative environment. Our initial results indicate that the optical properties of sand dominate the resultant signature for the buried and flush-buried cases. For surface landmines, the polarization results are dependent on the depth of the soil coating. Therefore, a spectral phenomenological model for the polarization signature of the combined sand-landmine system was developed for the infrared band (mid-wave and long-wave infrared) to study these issues in more detail. The modeling paradigm centered on a radiative transfer approach coupled with heat transfer results to account for incident and emitted radiation simultaneously. This paper will present a description of the physics-based model for the spectral polarization signatures of buried, flush, and surface land mines.

We propose a novel Stokes vector imagery system that is capable of determining the full Stokes vector of each pixel in an image simultaneously without any movable parts or modulation. The proposed system creates four channels by utilizing a fixed and rugged lenslet array to produce an exact replica of the incident image through each channel. Analysis of the instrument matrix singularities is discussed. Since all Stokes vector images are determined simultaneously, the processing speed of such a system is high, makes it very attractive for several important applications. Furthermore, the proposed system is expected to reduce several errors associated with conventional imaging polarimeters that employ movable parts.

Combining active multispectral and polarimetric imaging significantly enhances detection capability of very low contrast targets through the control of both the polarization state and the wavelength of the illumination light. However, increasing the operation range of the imaging system relies on the use of coherent sources, such as lasers and optical parametric oscillators, to illuminate the scene, leading to a dramatic decrease of the image quality due mainly to speckle noise. In order to investigate the benefits and drawbacks brought by coherent illumination, a preliminary laboratory demonstrator of an active multispectral polarimetric imager has been designed to operate with both polarized natural light and coherent sources. The orthogonal state contrast images recorded at different wavelengths in both configurations (coherent and non-coherent) clearly demonstrate the benefits of using active illumination of the scene to discriminate between real and fake targets and also to reveal very low contrast objects. Noise characteristics of polarimetric images under coherent illumination are also investigated. In particular the study of noise statistics of recorded images shows that the actual distribution of noise is log-normal. As a result, the so-called "natural" representation of the polarimetric image offers important advantages in terms of image processing. Indeed, if the intensity image is perturbed with multiplicative noise, the noise in the image with natural representation has uniform variance and is quasi-gaussian. The potential increase of target detection performance brought by properly processing the active polarimetric image is illustrated on a very low contrast scene.

Development of reliable imaging polarimeters and the models that predict their performance is dependent on the ability to assess their accuracy. Field tests frequently result in contradictory data and laboratory measurements are often not representative of materials in the field. To address these concerns, we have built a device with which the calibration of imaging polarimeters (both stationary and moving) can be verified and the polarimetric properties of materials in the field can be measured with accuracy. The device is a handheld, non-imaging polarimeter that is capable of highly calibrated phenomenology measurements in both the lab and field. Multiple optical heads enable monitoring of samples from a variety of angles in order to characterize polarimetric signatures as a function of source, sample, and sensor geometry. The device may also be used in unattended diurnal monitoring of polarimetric signatures of the sky, backgrounds, and targets of interest, providing a correlation between observed polarization phenomenology and weather conditions. The handheld device and the associated data acquisition system is small and portable enough that it can be taken to the field readily and is simple enough that calibration and system performance is predictable and verifiable. In this paper, we describe the design and performance of the non-imaging handheld polarimeter, performance specifications, and measurement results to date.

The Non-Scanning Computed Tomography Imaging Spectropolarimeter (NS-CTISP) is a novel imaging spectropolarimeter developed for snapshot imaging and collection of the four Stokes hyper-cubes. The patent pending NS-CTISP builds upon the computed tomography concepts proven in earlier imaging spectrometers and imaging spectropolarimeters by utilizing a combination of two-dimensional spectral dispersion and division of aperture polarization analysis to acquire all data necessary to estimate the Stokes hyper-cubes without scanning in spatial, spectral or polarimetric dimensions. A custom quadrant polarization analyzer and tetrahedron prism are used in tandem to perform aperture division and polarization analysis on each of the four pupils. Equations are developed which link the computed tomography spectral reconstruction and polarimetric calibration allowing the determination of Stokes vectors at each waveband. NS-CTISP is spectrally and polarimetrically calibrated by measuring the system’s spectral and polarimetric response to a radiating object having a small spatial extent, limited spectral bandwidth and a pure polarization state using a monochromator, fiber and polarization state generator. Iterative techniques are used to solve a linear imaging equation, where the solution provides input into the polarization calibration. Preliminary estimates of calibration fundamental accuracy indicates the average Stokes parameter error for central wavebands is on the order of 1.4-1.8 percent. A simple pseudo-object is created using two different fiber images each having a different polarization and reconstructed with average Stokes parameter error ranging from 0.7% to 3.2%. A suite of data acquisition, reconstruction and display programs have been developed in Interactive Data Language (IDL) to support NS-CTISP development.

We present adaptations of the channelled spectropolarimetry technique, a method which allows both spectral and polarization information to be captured in a single integration period. The first adaptation uses a mathematical decomposition of the system matrix, which is then modified for imaging spectropolarimetry; the second adaptation is applied first to a single-point and then to an imaging system, for which we also show applications and measurements from experimental work.

Over the past 25 years, classical optical interferometry (OIC) has become a mature discipline within astronomy. More recently, theoretical work has begun on optical interferometric polarimetry (OIP). Such analyses include adapting modern polarization mathematics from radio astronomy, modeling polarization effects in optical interferometers, and inventing observables with their corresponding output vectors. In this paper, I will demonstrate how OIP may be used to obtain significant results from relatively simple sources, such as spherically symmetric stellar atmospheres, spotted stars, and scattering envelopes.