Underwater Image Recovery Using Low-Frequency Filtering and Polarization Imaging Modeling

Aquatic medium attenuation results in reduced image contrast and detail ambiguity, constraining implementation of sophisticated vision tasks including object identification. The polarization imaging technique enhances multidimensional image information through measurement of degree of polarization (DoP) and angle of polarization (AoP) parameters. We propose an innovative underwater polarization imaging model derived from light wave polarization properties in aquatic environments, implementing image restoration via frequency-domain Gaussian low-pass filtering to accurately estimate background illumination. Empirical findings reveal the approach's effectiveness in detail enhancement and contrast improvement, facilitating performance optimization for various underwater visual processing tasks.