Infrared image super-resolution reconstruction based on residual fast fourier transform

Infrared images have been widely used in military, civilian, and industrial fields. Due to the inherent limitations of sensors, infrared images usually have some disadvantages such as low resolution and blurred texture details. How to improve the resolution of infrared images without changing hardware devices has become a current research hotspot. Aiming at the problem that the existing infrared image super-resolution reconstruction methods do not utilize the image details sufficiently, this paper proposes an infrared image super-resolution reconstruction algorithm based on residual fast fourier transform(ISRRFT). The image features are extracted in both spatial domain and frequency domain, so as to fully extract the high-frequency and low-frequency information components of infrared images and improve the quality of reconstructed infrared images In addition, the learnable fast fourier transform loss function has been introduced, which is used in conjunction with the L1 loss function to calculate the loss from both the spatial and frequency domains, thus better optimizing the model parameters. The test results on three test sets, SR1280, IRData and Infrared20, show that the proposed algorithm has an optimal performance in terms of both objective and subjective evaluation metrics compared to current representative algorithms. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2024.