Image Dehazing via Multilayer Detail Enhancement and Saturation Compensation

Images taken in hazy conditions often suffer from severe quality degradation, such as detail loss, low contrast, and color cast. To solve these problems, this paper proposed a naturalness preserved dehazing method. First, a hue histogram distribution prior is used to determine if the input image is color- casted, and a simple color correction method is provided to restore colors. Second, a multilayer detail enhancement unit is applied to sharpen the intensity channel, and an adaptive local gamma correction is used to compensate saturation. Dehazed results can be achieved by converting back to RGB. Extensive experimental results indicate that the proposed method can effectively improve the visibility of hazy images, bring out the details without undesirable halo effects, and reproduce vivid color. The assessment results demonstrate that our method is capable of achieving outstanding dehazing performance, compared with some state-of-the-art methods.