MvKSR: Multi-View Knowledge-Guided Scene Recovery for Hazy and Rainy Degradation

Wenyu Xu, Dong Yang, Yuan Gao, Yuxu Lu, Jingming Zhang, Yu Guo

Abstract: High-quality imaging is essential for effective safety supervision and intelligent deployment in vision-based measurement systems (VMS). It allows for accurate and comprehensive monitoring of operations, enabling the timely detection of potential hazards and efficient management. However, adverse weather conditions, such as atmospheric haziness and precipitation, can greatly affect the quality of imaging. The degradation is evident in the form of image blur and reduced contrast, increasing the likelihood of incorrect assessments and interpretations in VMS. To address the challenge of restoring degraded images in hazy and rainy conditions, this article proposes a novel multiview knowledge-guided scene recovery network (termed MvKSR). Specifically, guided filtering (GF) is performed on the degraded image to separate high/low-frequency components. Subsequently, an encoder–decoder-based multiview feature coarse extraction module (MCE) is used to coarsely extract features from different views of the degraded image. The multiview feature fine fusion module (MFF) will learn and infer the restoration of degraded images through cross supervision under different views. Extensive experimental results demonstrate that MvKSR outperforms other state-of-the-art methods in terms of efficiency and stability for restoring degraded scenarios, and can better serve the needs of advanced vision tasks in VMS. The source code is available at https://github.com/LouisYuxuLu/MvKSR.