Single Image Dehazing from Repeated Averaging Filters

In this paper, we present a method of Repeated Averaging Filters for estimating the atmospheric light from a single hazy image, which further contributes to better radiance recovery. Existing methods of dehazing are suffering from the problem of halo artifacts in the final output image after dehazing. For this purpose, an averaged channel is obtained from a single image by the repeated averaging filters via integral images which provides a faster and efficient way for removing halo artifacts. The proposed method of dehazing reveals competitive results in regards to qualitative, quantitative and computational analysis and out-performs many previous states of the art techniques.