The paper introduces a novel end-to-end single-image dehazing method called Densely Connected Pyramid Dehazing Network (DCPDN). DCPDN jointly learns the transmission map, atmospheric light, and dehazed image by embedding the atmospheric scattering model into the network. It features a densely connected encoder-decoder structure with multi-level pyramid pooling to estimate the transmission map, and a U-net for atmospheric light estimation. A joint discriminator based on GANs is used to ensure the quality of the estimated transmission map and dehazed image. The proposed method is optimized using a new edge-preserving loss function to preserve sharp edges. Extensive experiments on synthetic and real-world datasets demonstrate significant improvements over state-of-the-art methods.The paper introduces a novel end-to-end single-image dehazing method called Densely Connected Pyramid Dehazing Network (DCPDN). DCPDN jointly learns the transmission map, atmospheric light, and dehazed image by embedding the atmospheric scattering model into the network. It features a densely connected encoder-decoder structure with multi-level pyramid pooling to estimate the transmission map, and a U-net for atmospheric light estimation. A joint discriminator based on GANs is used to ensure the quality of the estimated transmission map and dehazed image. The proposed method is optimized using a new edge-preserving loss function to preserve sharp edges. Extensive experiments on synthetic and real-world datasets demonstrate significant improvements over state-of-the-art methods.