Near-Infrared Spectroscopy (NIR) is valuable for getting thorough information the human eye cannot capture in agricultural operations and investigations. A special multispectral camera is required to get NIR spectral data, which provides low-resolution images compared to RGB (Red–Green–Blue) data. NIR spectral data contains low contrast and no colour information making it difficult to identify and match features in the NIR domain. We require more features for image stitching to achieve large FoV (Field of View) NIR images. We use Cycle Generative Adversarial Network (CycleGAN) for image-to-image translation to synthesise an RGB image from a NIR spectral band image to take advantage of both RGB and NIR bands for the feature extraction. Features are extracted from NIR and synthetic RGB images. Then, we stitch the different NIR views using concatenated features of both domains to create large Field of View (FoV) NIR images. This augmented feature set yields superior stitching results compared to NIR features alone. The effectiveness of the proposed framework is demonstrated through experiments, showcasing the improvements in stitching quality.