Polarization dehazing method based on spatial frequency division and fusion for a far-field and dense hazy image

This is the authors' abstract. We don't add key points for this paper.

Polarization dehazing technology is effective in imaging through scattering media because of additional information different from the light intensity and spectrum. However, the existing methods relying on the manual choice of bias factor are non-universal in different imaging conditions. In addition, these methods are not suitable for dense scenes with long distances. Aiming at the dehazing application requirements in far-field and dense hazy weather, a polarization dehazing method based on spatial frequency division and fusion (SFDF) is proposed in this paper. In addition, we optimize the interpolation process before dehazing so that the spatial resolution can be maintained without the noise influence. The experimental results indicate that the proposed method outperforms the existing schemes in dense hazy weather more than kilometer distances. Furthermore, we discuss that the effects of bias factors only act on the low-frequency parts of the polarization images, and their influence is greatly weakened after being fused with the high-frequency parts. This robust advantage without manual intervention causes the proposed SFDF method to have a broader prospect in practical application scenarios.