Kodlayıcı-kod çözücü sinir ağı ile kızılötesi ve görünür spektrum görüntülerde füzyon

The image fusion aims to gather all important information from the source images into a single image. While the data is reduced, the fusion image has a high spatial and spectral resolution. It includes more informative and complete information. In this work, we reviewed state-of-the-art methods in the infrared and visible spectrum image fusion literature and we present a novel deep learning-based solution. Our proposed method is inspired by encoder-decoder network U-Net architecture. Furthermore, we analyzed the fusion quality measurement metrics. We integrated fusion quality measurements into our proposed method's training step. In this way, we achieved superior performance. The analysis is performed qualitatively and quantitatively on TNO and VIFB datasets. The proposed method is compared with state-of-the-art methods and detailed experiments are conducted. It shows the best performance among deep learning-based methods. Project codes can be found at https://github.com/ferhatcan/pyFusionSR.