| Title |
Image Dehazing and Attenuation Coefficient Estimation using Depth Estimation in a Single Image |
| DOI |
https://doi.org/10.5573/ieie.2023.60.6.65 |
| Keywords |
Image dehazing; Deep learning; Depth estimation; Airlight estimation; Attenuation coefficient |
| Abstract |
Outdoor images can provide distorted information due to the presence of the haze effect, which can cause performance degradation in various computer vision applications. Numerous researches have been performed on image dehazing. Image processing and deep learning methods have used typically the atmospheric scattering model for image dehazing. This paper proposes efficient methods for image dehazing that involves quick search for the optimal attenuation coefficient of a hazy input image. Deep learning models are used to estimate the depth information and atmospheric light information of the single input image, and image dehazing is then performed based on the atmospheric scattering model with the optimal attenuation coefficient. This paper proposes three search methods, namely binary search, quartile search, and full search, to enable the fast search for the optimal attenuation coefficient. The experiment results compare the three search methods based on metrics such as PSNR, SSIM, entropy, and FRFSIM with previous methods, and show that the proposed methods provide relatively good performance and reduced execution time. |