Single Image Dehazing Based on Bright Channels Prior Compensation

Abstract

Image taken in hazy environment usually have low contrast, low visibility, and blurred details. To solve this problem, we proposed a method to remove haze from a single image. Estimation of atmospheric light and transmission are the key steps of image dehazing. Due to the dark channel prior fails in large areas of sky or dense fog, a transmission estimation approach based on bright channel prior compensation is put forward, which can effectively avoid the problem of color distortion of dehazed image, and the estimated transmission is more accurate after the guide image filtering. At the same time, when the image is processed in the dark channel, the atmospheric light is the global atmospheric light, that is, the atmosphere of each pixel is the same. This approach proposes a method of getting the final atmospheric light via local atmospheric light estimation and global atmospheric light estimation, which solves the problem of atmospheric light is not the same for each pixel and the brightness of the restored image is low. Experiments show that the method in this paper can effectively remove haze, enhance image contrast and details. Compared with other algorithms on real-world images, our algorithm shows favorable performance for haze removal.