Multi-Pyramid Dehazing Network with Residue Channel Enhancement
Konferenz: CAIBDA 2022 - 2nd International Conference on Artificial Intelligence, Big Data and Algorithms
17.06.2022 - 19.06.2022 in Nanjing, China
Tagungsband: CAIBDA 2022
Seiten: 4Sprache: EnglischTyp: PDF
Autoren:
Hang, Ruishan; Tang, Zhibo (School of Information Engineering, Zhejiang Ocean University, Zhoushan, China)
Gu, Shenming; Tan, Xiaoqiu (School of Information Engineering, Zhejiang Ocean University, Zhoushan, China & Key Laboratory of Oceanographic Big Data Mining & Application of Zhejiang Province, Zhoushan, China)
Inhalt:
Single image dehazing is to remove haze from degraded images and recover clean scenes. The researches of dehazing have academic significance and application value. In this paper, we design a Multi-Pyramid Dehazing Network (MPDN) with Residue Channel Enhancement (RCE). A RCE module was designed to help the network model remove a large amount of homogeneous haze. To further improve the effect of eliminating heterogeneous haze, the Pyramid Feature Extraction (PFE) module is used. The two modules complement each other to achieve excellent haze removal effect. Finally, we do a lot of experiments to verify the model, and compare it with the state-of-the-art (SOTA) method on the benchmark datasets.