A Novel Nighttime Dehazing Model Integrating Retinex Algorithm and Atmospheric Scattering Model

Abstract

To solve the problem of low contrast and serious loss of detail information after dehazing directly using the atmospheric scattering model. This paper proposes a novel night image defogging model that combines the Retinex and the atmospheric scattering models. This model incorporates the best of both algorithms. Retinex can enhance the contrast of the image and the detailed information of the image. The atmospheric scattering model can clearly describe the imaging principle of the haze map, and restore the haze-free image from the root. In qualitative and quantitative experiments, this paper compares four methods on public datasets. This algorithm’s average image information entropy processed by this algorithm is 6.995, the average peak signal-to-noise ratio is 19.340, and the average structural similarity is 0.766. Compared with the other four algorithms, the comparative test results show that the algorithm in this paper can effectively remove the haze in the nighttime image, restore the detailed information of the image, and promote the progress of the algorithm.