Image Dehazing With Polarization Boundary Constraints of Transmission

We present an image dehazing approach using polarization boundary constraints of transmission (Pol-BCT) in this work. First starting from the polarization performance of the sensor, we derive the calculation methods of polarized airlight and infinite airlight by introducing Muller matrix; this method dramatically reduces noise in image dehazing. Then we propose using degree of polarization (DoP) of airlight to construct polarization transmission and reconstruct the atmospheric scattering model (ASM); the constructed ASM is more suitable for dehazing in most scenes. We propose boundary constraint conditions based on polarization to refine this transmission, which can retrieve the main profile information in various scenes. To evaluate the proposed Pol-BCT, we create a polarimetric hazy image datasets. Experimental findings show that the suggested approach outperforms previous dehazing algorithms in both qualitative and quantitative evaluations.