SGXAI_FHO: Spiking Graph Explainable Artificial Intelligence with Fire Hawk Optimizer-based fusion of infrared and visible images under low-light conditions

The traditional methods for fusing images acquired under low-light conditions, in many cases, are unable to retain sufficient integrity of structural and textural features, resulting in a distorted and degraded ability to detect features within a scene. This paper presents a novel approach to overcome these issues by using Spiking Graph Explainable Artificial Intelligence (SGXAI) with Fire Hawk Optimization (FHO) to effectively fuse infrared and visible image data. This new SGXAI_FHO framework uses a combination of Gaussian smoothing algorithms to reduce image noise, brightness, and contrast enhancement techniques on visible images to improve the image quality perceived by the human eye (the observer), as well as cross-modal attention (CMA) mechanisms to extract complementary features from the two input modalities and to prioritize these features during the fusion process. The results of this experimentation provide compelling evidence of the effectiveness and accuracy of the proposed framework, achieving an F1-score of 99.81%, precision of 99.46%, recall of 99.72%, and accuracy of 99.56%. These results demonstrate that the SGXAI_FHO framework will provide a robust, reliable, and interpretable approach for low-light image fusion and detection applications and has substantial potential for real-world applications.