34 / 2017-04-30 13:19:20

Image Enhancement Algorithm for Highway Tunnel Based on Imaging Model Estimation

Abstract—Highway tunnel environment is dim, with the interaction of multi-color light sources, which forms a light haze. Such a complicated scene makes the image blurred and arouses difficulties in the image enhancement. According to the law of haze imaging, a haze removal algorithm for nighttime scene is applied to enhance the tunnel image in this paper. First, the global atmospheric light is estimated as the lower limit of atmospheric light, and the shortest Euclidean distance in the space of brightness is sought as the optimization. Such two results are combined to estimate the local atmospheric light. The lower limit of tunnel local transmission image is then determined by finding the maximum information entropy as the optimization. Tunnel local transmission is estimated based on the weighted smoothness. Finally, the tunnel image is reconstructed after extinction process. The experimental results validate the effectiveness of the proposed algorithm for tunnel image enhancement.