LAPNet

LAP-Net: Level-Aware Progressive Network for Image Dehazing

An overview of our method. We use the cascaded hourglass units to construct t-net, which is used for estimating the transmission map stage by stage with the supervision of different haze levels. Meanwhile, a residual dense pooling network serves as the A-net to learn the atmospheric light. The predictions of t-net and A-net are sent to the restoration layer to generate progressive restorations. Then we send the restored images of each stage as the input to the I-net for integration. The I-net cooperates with a hierarchical integration scheme that selects the clear regions of each stage and weights them with the guidance of haze level to restore the ﬁnal haze-free image.

Abstract

In this paper, we propose a level-aware progressive network (LAP-Net) for single image dehazing. Unlike previous multi-stage algorithms that generally learn in a coarse-tofine fashion, each stage of LAP-Net learns different levels of haze with different supervision. Then the network can progressively learn the gradually aggravating haze. With this design, each stage can focus on a region with specific haze level and restore clear details. To effectively fuse the results of varying haze levels at different stages, we develop an adaptive integration strategy to yield the final dehazed image. This strategy is achieved by a hierarchical integration scheme, which is in cooperation with the memory network and the domain knowledge of dehazing to highlight the best-restored regions of each stage. Extensive experiments on both real-world images and two dehazing benchmarks validate the effectiveness of our proposed method.