Due to hardware limitations, multispectral videos often exhibit significantly lower resolution compared to standard color videos. These videos capture images in multiple bands of the electromagnetic spectrum, providing valuable additional information that is not available in traditional RGB images. This paper proposes a Bayesian approach to estimate super resolved images from low-resolution spectral videos. We consider adjacent frames from a video sequence to provide a super-resolution image at a time. We include in our proposal the motion between adjacent frames and unlikely to the literature, we estimate the blur and noise while reconstructing the higher resolution image. Experimental results on spectral videos demonstrate the effectiveness of our approach in producing high-quality super resolved images.