Image defogging method combining variational model and alternating least squares

Traditional image dehazing methods based on physical model do not consider the coupling of different color channels, which will produce artifacts during dehazing. To solve this problem, an improved image fog removal method based on total variational model is proposed. First of all, in order to improve the accuracy of atmospheric light estimation, dark channel principle is used to calculate atmospheric light, which can effectively improve the effect of fog removal. Secondly, based on the atmospheric scattering model, alternate least-squares processing is carried out on the de-fogging image, and the transmission of the three RGB color channels is coupled to effectively remove the artifacts existing in the image. The experiment is conducted in the LIVE image de-fogging database to compare with the classical de-fogging method. The experimental results show that the proposed algorithm not only reduces the artifacts in images, but also increases the peak signal-to-noise ratio and structural similarity by 2.1 and 0.5 respectively, and improves the quality evaluation of unreferenced images by 1.7, which proves the effectiveness of the proposed algorithm.