Infrared images can provide prominent targets based on the radiation difference, making them suitable for use in all day and night conditions. On the other hand, visible images can offer texture details with high spatial resolution. Infrared and visible image fusion is promising to achieve the best of both. Conventional frequency or spatial multi-scale transformation methods are good at preserving image details. Deep learning-based methods become more and more popular in image fusion because they can preserve high-level semantic features. To tackle the challenge in extracting and fusing cross-modality and cross-domain information, we propose a Spatial-Frequency Collaborative Fusion (SFCFusion) framework that effectively fuses spatial and frequency information in the feature space. In the frequency domain, source images are decomposed into base and detail layers with existing frequency decomposition methods. In the spatial domain, a kernel-based saliency generation module is designed to preserve spatial region-level structural information. A deep learning-based encoder is employed to extract features from the source images, decomposed images and saliency maps. In the shared feature space, we achieve cross-modality spatial-frequency collaborative fusion through our proposed adaptive fusion scheme. We have conducted experiments to compare our SFCFusion with both conventional and deep learning approaches on TNO, LLVIP and M ³ FD datasets. The qualitative and quantitative evaluation results demonstrate the effectiveness of our SFCFusion. We have further demonstrated the superiority of our SFCFusion in the downstream detection task. Our code will be available at https://github.com/ChenHanrui430/SFCFusion.

High-quality infrared images are widely used in various vision tasks. However, infrared focal plane arrays (IRFPAs) suffer from material and process limitations, resulting in nonuniform pixel response and severe spatial fixed pattern noise (FPN), which results in degraded images. To address the problem that existing infrared image nonuniformity correction (NUC) methods only focus on stripe noise removal but ignore environmental noise, we propose a new multistage single-frame NUC processing pipeline. First, for the low-rank property of infrared degraded images and the higher-order directional priori information of stripe noise, we construct an objective function that combines a higher-order gradient total variation model with a diagonal kernel paradigm as a constraint term. Multiple subproblems are solved by the alternating direction method of multipliers (ADMMs) to obtain the recovered image after the removal of the stripe. Then, we incorporate the environmental noise into the pipeline by low-rank matrix approximation and employ singular-valued patch decomposition to efficiently separate the clean image from the noise. Extensive experiments are conducted on existing methods on real and simulated datasets, and the potential performance of the proposed method is verified in qualitative and quantitative evaluations. The code and datasets can be obtained at https://github.com/ImageVisioner/InfraredNUC.