Image defogging algorithm based on sky area recognition

Abstract

Foggy weather severely reduces the image quality and increases the difficulty of obtaining aerial image information from UAVs. In this paper, an image defogging algorithm based on sky region recognition is proposed, which solves the problems such as inaccurate estimation of sky region transmittance, halo, and color distortion after restoration. Firstly, this paper uses the improved Otsu algorithm to divide the image into sky region and non-sky region, and obtains the atmospheric light value by identifying the sky region. Secondly, the light transmittance of non-sky region is obtained by using the adaptive dark channel constructed by local information entropy, and the light transmittance of sky is obtained by using the improved linear model. Then, the overall transmittance is fused and refined. Finally, Leigh obtained the initial recovery image in the atmospheric scattering model, and the final image was obtained by adaptive color adjustment. The experimental results show that the algorithm has a good defogging effect, and both subjective and objective evaluation is better than the related algorithms.