Network-Assisted Super-Resolution Imaging of Magnetic Fields Under Noisy Conditions

High-resolution (HR) magnetic field signals are crucial for system monitoring and fault diagnosis. However, collecting HR magnetic field data often requires a significant amount of time and is susceptible to interference from the surrounding environment. To address these issues, an experimental collection platform is designed to collect low-resolution (LR) signals, which are transformed into LR images, and then a novel network for image super-resolution (SR) reconstruction is developed to obtain the HR signals. Moreover, to facilitate the recovery of HR data while removing the disturbance, a multifeature fusion image SR reconstruction network RSGLEAN is proposed, which introduces the residual Swin Transformer block (RSTB) as a feature extraction module to effectively filter out noise interference and the deep feature extraction module (RRDB) integrated with an attention mechanism to further extract relevant information for image SR. Experimental results demonstrate that the proposed RSGLEAN network outperforms other SR methods in terms of peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM) under interference conditions, which justifies the capability and efficiency of our method for collecting HR magnetic field signals. This method offers a promising solution for high-precision magnetic field signal collection and novel ideas for the development of new sensors.Index Terms— Deep neural network, image super-resolution (SR) reconstruction, magnetic field imaging, magnetic field signal detection.