Real-Time Restoration of Power Distribution Scenario Monitoring Images Based on Mamba

To ensure stable operation of the power distribution equipment, it is necessary to monitor the operational status of the equipment in real time. Video surveillance, which records image information from distribution scenarios, is a common method to monitor these devices. However, the quality of surveillance images can degrade due to interference from imaging conditions. To ensure the quality of real-time monitoring images, it is crucial to promptly detect image quality degradation and perform real-time restoration of the images. Unlike traditional non-causal restoration tasks based on stored videos, real-time image restoration requires the construction of a causal system. This system restores the current un-degraded image based on historical and current degraded images. For this purpose, we propose a real-time restoration model for Mamba-based surveillance images. Our method performs well in the power distribution scenario monitoring image dataset we constructed, achieving a PSNR of 33.60.).