Polarization image defogging based on the dark channel principle and multi-scale image fusion

To enhance the recovery capability of the polarization defogging algorithm for foggy scenes, a polarization image defogging algorithm founded upon the dark channel principle and multi-scale fusion is put forth. Initially, the atmospheric light is estimated based on the polarization information. This is followed by the estimation of atmospheric transmittance, which is achieved by combining the dark channel principle with the multi-scale image fusion technique. Finally, the atmospheric scattering model is constructed to obtain the defogged image. The experimental results demonstrate that the proposed algorithm exhibits excellent performance, achieving a high correlation coefficient of relevance and low CIEDE2000 color difference values in simulated fog environments of varying concentrations. The defogged image is improved by 59.68 times in terms of contrast, and the information entropy is increased by 11.38%. In a real outdoor fog scene, the defogged image generated by this algorithm showed a contrast improvement of 25.24 times and an information entropy enhancement of 8.32%. The presented algorithm defogs images more effectively and naturally than existing algorithms, enabling the recovery of detailed information from fog-obscured scenes.