The research on de-raining algorithm for degraded images of transmission line wildfire smoke based on the multistage progressive network

Monitoring and early warning of wildfire smoke events is essential for ensuring the stable operation of transmission lines. Currently, the monitoring methods deployed in transmission corridors primarily target clean and simple environmental conditions, exhibiting poor weather resistance. As a result, they are unable to effectively identify and monitor wildfire smoke under rainy weather conditions, which increases the safety risks associated with power grid operations. To address this issue, inspired by the multi-stage progressive de-raining algorithm, a rain removal method based on the multi-stage progressive network is proposed for rainy images of wildfire smoke in transmission corridors. The method models the rain removal process into three sub-processes: rain streak elimination, rain artifact removal, and image structure restoration. Each sub-process is handled by an independent subnetwork, with a supervised attention module progressively calibrating the predictions of each subnetwork, ultimately achieving the removal of rain streaks. Quantitative and qualitative comparison experiments with commonly used rain removal algorithms demonstrate that the method proposed not only effectively removes rain streaks but also retains the complete details of the transmission corridor and smoke images, providing a technical foundation for power visual technology in the operation and management of power equipment under extreme weather conditions.