In several manuscripts, text is obscured due to glued together leaves, making it difficult or impossible to read. In this context, this study explores the use of reflectance and transmittance imaging spectroscopy (RIS and TIS) in the visible and near-infrared range (400–1000 nm) to recover hidden texts. The method is applied to two cases of medieval Frisian legal codes from the Richthofen Collection, where we employed Non-Negative Matrix Factorization (NMF) for the analysis of single and spectrally fused datasets integrating both RIS and TIS. We further integrated spatial stitching of adjacent areas to enhance spatial resolution of the images. Our results demonstrate that factorization algorithms perform well on fused datasets, with spectral fusion proving essential in complex cases where individual analyses fail to clearly reveal hidden text.