Images taken in scattering media, such as haze, fog, and underwater, often look faded and lack contrast. The degradation is different for every pixel and depends on the distance of the scene point from the camera. For terrestrial images, this dependency is expressed in the transmission coefficients, that control the scene attenuation and amount of haze in every pixel. Previous methods solve the single image dehazing problem using various patch-based priors. We, on the other hand, propose an algorithm based on a new and non-local prior. The algorithm relies on the assumption that colors of a haze-free image are well approximated by a few hundred distinct colors, that form tight clusters in RGB space. Our key observation is that pixels in a given cluster are often non-local, i.e., they are spread over the entire image plane and are located at different distances from the camera. In the presence of haze these varying distances translate to different transmission coefficients. Therefore, each color cluster in the clear image becomes a line in RGB space, that we term a haze-line. Using these haze-lines, our algorithm recovers both the distance map and the haze-free image. We show how to expand the model to restore the colors of underwater images, by incorporating spectral dependency of the attenuation coefficients. The algorithm is linear in the size of the image, deterministic and requires no training. It performs well on a wide variety of images and is competitive with other state-of-the-art methods.
Dana Berman is a PhD candidate at Tel Aviv University, under the supervision of professors Shai Avidan and Tali Treibitz. She holds an M.Sc. in Electrical Engineering from Tel Aviv University and a B.Sc. in Electrical Engineering and Physics from the Technion. Her research focuses on color and contrast restoration of images taken in bad weather conditions and under water.