Adaptive polarimetric image representation for contrast optimization of a polarized beacon through fog

TL;DR

This paper introduces an adaptive linear representation method for polarimetric images that maximizes contrast of a polarized beacon in foggy conditions, outperforming simple difference images and validated by experiments and noise modeling.

Contribution

The paper proposes a novel adaptive polarimetric image representation that optimally enhances contrast in obscured weather conditions, based on noise correlation analysis.

Findings

Adaptive representation outperforms simple difference images in contrast enhancement.

Experimental results align with the correlated Gaussian noise model.

Contrast gain is analytically derived and experimentally validated.

Abstract

We present a contrast-maximizing optimal linear representation of polarimetric images obtained from a snapshot polarimetric camera for enhanced vision of a polarized light source in obscured weather conditions (fog, haze, cloud) over long distances (above 1 km). We quantitatively compare the gain in contrast obtained by different linear representations of the experimental polarimetric images taken during rapidly varying foggy conditions. It is shown that the adaptive image representation that depends on the correlation in background noise fluctuations in the two polarimetric images provides an optimal contrast enhancement over all weather conditions as opposed to a simple difference image which underperforms during low visibility conditions. Finally, we derive the analytic expression of the gain in contrast obtained with this optimal representation and show that the experimental results are in agreement with the assumed correlated Gaussian noise model.