Modified Degree of Polarization Function for Rough Metallic Surface Parameter Estimation Based on Multispectral Polarimetric Measurement

Abstract

In order to recognize different materials from passive polarimetric measurements specifically, a model that can describe the distribution of polarization energy scattered from the target surface is needed for estimating the parameters of interest (complex refractive index n, k and surface roughness ) of targets. This paper presents a new method for targets parameters estimation. The presented method modifies the degree of linear polarization (DOLP) model by introducing the Lorentz-Drude dispersion constants which do not vary with wavelength as well as the diffuse component of reflection related to the surface roughness to the DOLP equation. This solves the problem of difficulty and inaccuracy caused by multi-angle passive polarimetric measurements used by previous works and helps to improving the parameters estimation accuracy for rough targets. The Monte Carlo simulations result illustrate that the presented method is robust to Gaussian noise. Moreover, the experimental results of rough aluminium plate and rough copper plate validate the effectiveness of the presented model. The errors of estimated n are about 8% for aluminium plate and 9% for copper plate. The errors of estimated k are less than 4% for aluminium plate and no more than 5% for copper plate. The errors of estimated surface roughness  are about 6.8% and about 6% for aluminium plate and copper plate respectively. The high estimation accuracy and less detection angle requirement achieved by the presented method are significant for material characterization in passive remote sensing application.

Polarization spatial and semantic learning lightweight network for underwater salient object detection

“…Polarization imaging has been well documented for its excellent recognition capabilities in applications, such as object detection, 26 transparent object detection, 27 and background segmentation 28 . Polarization images can obtain physical properties of objects that cannot be detected by conventional imaging, such as degree of polarization, roughness, and other special properties 29 . In underwater scenes, due to the interaction of a light wave with an object, the polarization information can differ significantly depending on the object material, surface roughness, geometry, and so on, which enhances the detection of the target compared with traditional optical images.…”