Contents I. Introduction
Remote sensing images provide a wealth of spatial and geographic information and are widely used for forestry, meteorology, hydrology, and military. However, they are easily degraded by atmospheric scattering due to suspended particles in the atmosphere, such as haze, fog, and mist, which will reduce their application value to a great extent. Therefore, many researchers studied on dehazing to improve image quality. He et al. [1] proposed a dark channel prior to recover a high-quality haze-free outdoor image. Ancuti et al. [2] introduced a fusion-based strategy to single image dehazing, which achieves satisfactory results and can be used for real-time applications. Makarau et al. [3] restored the haze-free remote-sensing image through subtracting its corresponding haze thickness map, which is constructed by searching dark objects locally in the whole image. For some certain applications, remote-sensing images from Google Earth are one of the important data sources, such as target detection [4], [5] and classification [6], [7]. This kind of images are also influenced by hazy condition; therefore, two fast dehazing methods based on dark channel prior [8] and deformed haze imaging model [9] are proposed, respectively, to remove haze from Google Earth images. Some researchers also studied on how to evaluate the dehazing effects. Fang et al. [10] proposed an evaluation metric combining the ascension of the contrast degree with the structural similarity to assess the image quality after dehazing.
