Kashif Usmani, University of Connecticut; Timothy O’Connor, University of Connecticut; Peter Marasco, Air Force Research Laboratory; Bahram Javidi, University of Connecticut
Keywords: Polarimetric imaging, three-dimensional integral imaging, degraded environments, Long Wave Infrared (LWIR) imaging.
Abstract:
We investigate the performance of two-dimensional (2D) and three-dimensional (3D) polarimetric imaging in low illumination conditions using both visible range and long-wave infrared (LWIR) range imaging sensors. Polarimetric imaging is an important imaging strategy within the fields of object visualization, object recognition, materials inspection, and materials classification. While relying on the light reflected from objects under study, polarimetric imaging can become challenging in degraded environments such as low illumination conditions or in the presence of partial occlusions. 3D integral imaging (InIm) is one prominent technique for improving scene visualization under such degraded conditions by using multiple perspectives of 2D elemental images of the scene to reconstruct the 3D image of the scene. This allows depth information to be extracted and improves the performance of low light image reconstruction due to being optimal in the maximum likelihood sense. Together, 3D polarimetric InIm can measure 3D polarimetric information of objects in photon-starved conditions and improve the performance of polarimetric imaging over conventional 2D techniques. Moreover, the reconstructed 3D integral image at a particular depth or range under extremely low light environments may provide a higher signal to noise ratio (SNR) as compared to the 2D images. In this paper, we present new data on the performance of visible range camera and long-wave infrared (LWIR) imaging systems in the extraction of polarimetric information of objects in photon starved conditions. The polarimetric information of the object can be extracted from the Stokes parameters and degree of polarization (DoP). The Stokes polarization parameters will be measured using a rotating polarizer in front of an image sensor and applied for the calculation of DoP image of scene. An LWIR wire grid polarizer and a linear polarizer will be used as the polarimetric objects in adverse environmental conditions for LWIR range and visible range imaging systems, respectively. Since the number of photons per pixel will be very low in case of visible range sensing system, the total variation (TV) denoising algorithm and mathematical restoration model will be applied to the visible DoP images to enhance the visualization. The mathematical model for signal restoration and TV denoising algorithm will be not applied in case of the LWIR imaging system because enough thermal photons will be present in the scene. For completeness, a quantitative assessment of these preprocessing steps will be included and will be shown to have no benefit in the case of the LWIR imaging system. A quantitative comparison on the performance of imaging systems in the extraction of polarimetric information for the object under adverse environmental conditions including low light conditions in terms of SNR is presented in this paper. We also show that a small polarimetric object is detectable in visible 3D polarimetric integral imaging while it is hardly detectable by the LWIR polarimetric imaging system. Finally, we derive theoretically the probability density function for the 2D and 3D DoP images and show a strong similarity between the theoretically derived distributions and the probability distribution functions found through experimental measurements. These results demonstrate significantly better performance in visible range sensing for polarimetric 3D integral imaging over conventional 2D imaging in degraded environments. The results also demonstrate the effectiveness of the chosen preprocessing methods for visible range sensing and show overall better performance by the 3D integral imaging visible range sensing system in comparison to the LWIR imaging system.
Date of Conference: September 14-17, 2021
Track: Optical Systems & Instrumentation