Imaging Spectropolarimetry research addresses the following hypothesis: Polarization-sensitive geometric phase holograms can be used to implement snapshot hyper-spectral polarization imagers with several performance advantages. We used numerical simulation to assess reconstruction performance, and showed a favorable comparison with prior work. We also found that several potentially optimum dispersion patterns, to be evaluated further, and identified favorable characteristics. Moreover, we found algorithmic technique to vastly improve SNR iterative convergence. The primary scientific contribution within the research is demonstration of working principle of the hyper-spectral polarization imaging with two PGs for lab-scale and natural objects. We also derived governing theory of operation and design principles, and developed an optimum calibration approach that is simple and generic, applicable to any PG-based imaging polarimeter. Although direct comparison of prior and our imaging system is nearly impossible, we showed bench top results that suggest substantially more accurate reconstruction spectrally and spatially for the full Stokes vector.

The overall purpose of the research is to make an in depth study of digital imaging systems as building and developing an imaging hyper-spectral polarimeter based on novel liquid crystal diffraction grating. The imaging system will allow us to measure spectral and polarization information of dynamic events with no moving or tunable part by using the liquid crystal diffraction gratings, also known as polarization gratings (PGs). Spatially varying birefringence patterns of PGs lead to the imaging spectrum (red to violet) from the angular dispersion and polarization separation (full Stokes parameters) from the order of diffractions. By using the multiple PGs with proper configurations, we can achieve full spectral and polarization separation on a single detection arrays such as CCD or CMOS.

There are different types of information that can be extracted from a scene. These include spatial information captured by detector (ex. normal cameras), spectral information retrieved by spectrometers, and polarization states achieved by polarimeters. Each one of these instruments shows a different way of identifying objects. For example, the spectral information specifies material components in a scene, whereas the polarization information tells us the surface feature, shape, and roughness with vector contents of the optical field. Measurements of the spectral and polarization information of the image are used in a various fields such as astronomy, remote sensing, atomic and molecular physics, and material characterization.

Next pictures show one example of the ability of polarization to defeat spectral camouflage. From the previous research we could know that manmade objects are source of emitted and reflected polarization while natural backgrounds are predominantly unpolarized. Even when the target is not spectrally camouflaged, polarization is used to improve target contrast. Applications of imaging spectropolarimetry are employed in several fields including geology, where they are used in remote sensing to locate and identify Al, Cu, Fe, Pb and quartz based on their polarized reflection spectrum. Astronomy is possibly the most prevalent application of spectropolarimetry. Spectral contents are used to classify the spectral signature and locations of distant stars, study the composition of planets, and characterize other celestial objects by their emitted and reflectred spectra, while the polarization contents are used to determine the structure of various types of nebulae and galaxy.

Hyper-Spectral Imaging Polarimetry
Most current methods to acquire spectral and polarimetric information need moving parts or modulation processes which lead to significant complexity or reduce sampling resolution. In this work, we presented a novel snapshot imaging spectropolarimeter based on anisotropic diffraction gratings known as polarization gratings. Using multiple PGs and waveplates, we can simultaneously acquire both spectrally dispersed and highly polarized diffractions of a scene on a single focal plane array. The PGs create chromatic dispersion (spectral information) patterns are linearly proportional to Stokes vectors (polarization information) embedded in a scene. PGs uniquely produce only three diffracted orders (0 and ±1), polarization independent zeroth-order, polarization sensitive first-orders that depend linearly with the Stokes parameters, and easily fabricated as polymer films suitable for visible to infrared wavelength operation.

We develop and test a system matrix for reconstructing the object information from this diffraction pattern. This matrix can be extended to various configurations containing several PGs. From a matrix representation, a continuous model can be converted to spatially and spectrally quantized model in discrete manner. Moreover, our imaging spectropolarimeter is calibrated with single wavelength light source (i.e. laser) and narrow-band color filters, and is experimentally demonstrated with scenes having spectral and polarization variation. This demonstrates an imaging spectropolarimeter approach, that reconstructs both screen generated scenes and outdoor objects. Reconstructed objects are sampled at 100 × 100 × 51 (x, y, λ) with 4 nm spectral resolution. The most significant advantage of our spectropolarimeter over other snapshot imaging systems is its capability to provide simultaneous acquisition of both spectral and polarization information at a higher resolution, and in a simpler and more compact way.

[1] Kim et al. Proc. SPIE 7086, 708603 (2008)
[2] Kim et al. Proc. SPIE 7672, 767208 (2010)
[3] Kim et al. Proc. SPIE 8364, in preparation (2012)