Degraded image restoration based on quadtree decomposition in scattering media

Image restoration has attracted the attention of many scientists due to the image is degraded by the bad weather (such as haze, smog, fog). Clear images provide a means for security surveillance, remote sensing and various military application to understand objective facts. Many dehaze methods have been proposed by the experts for image restoration, especially for image dehaze. The dark channel prior dehaze method is a typical image restoration method based on atmospheric physical model. This method is a kind of statistics of outdoor haze-free images, and it is a simple but effective remove haze from a single input image. However, this method fails to restore the sky region of degraded image, and it has a high computational cost associated with soft matting algorithm. To overcome these problems, we propose an image restoration method based on quadtree decomposition to restore the images degraded by scattering media. The proposed method uses the quadtree decomposition to find the sky region for the atmospheric light estimation. The transmission of sky region is improved by the proposed method to obtain an accurate global transmission of degraded image. The degraded image can be quickly restored by our proposed method without halo effect or color distortion. The proposed method will be helpful to the security surveillance, remote sensing and various military application et al.