Infrared and visible image fusion using GAN with fuzzy logic and Harris Hawks optimization

The fusion of visible and infrared images has gained great importance and attention with the emergence of machine learning techniques for applications such as surveillance and medical diagnosis. This study proposes a novel Fuzzy Generative Adversarial Network (FGAN) integrated with a Harris Hawks Optimization (HHO) algorithm to fuse visible and infrared images, addressing the limitations of existing approaches by dynamically optimizing a Mamdani-type fuzzy logic system within the generator. The innovation lies in employing HHO to tune fuzzy rules and output membership functions-using entropy, PSNR, and SSIM as training targets-alongside a Support Vector Machine (SVM) enhanced with Frechet Inception Distance (FID) for discriminator training, justifying the use of multiple models to achieve robust feature extraction and classification. Unlike traditional methods, this hybrid approach ensures superior adaptability to diverse image characteristics. This work achieves remarkable adaptability in infrared and visible picture fusion by incorporating a fuzzy logic system, optimized by HHO, within a GAN framework, therefore pioneering a revolutionary fusion paradigm. Superior visual quality and fusion precision are achieved by the suggested FGAN's unique integration of a Support Vector Machine and FID in the discriminator, which greatly improves differentiation accuracy. Experimental results on the TNO dataset using MATLAB demonstrate that the proposed FGAN outperforms state-of-the-art techniques, achieving a PSNR up to 55 dB, an SSIM up to 0.99, and SF lower than 10, reflecting enhanced visual quality and information retention. By providing high-quality fused images with better clarity and information retention, this method greatly improves remote sensing, surveillance, and medical diagnostics.