Local entropy based image reconstruction

This article presents a local entropy based image reconstruction algorithm that performs quite well in cases where there is distortion in an image. If the "wanted" image information is still available but distributed over two or more distinct images, the algorithm can collect the required information from the set of the images given. Instead of inferring pixel data information from the remainder of one single image, the algorithm provides a decision rule on what information from which one of the set of given images to actually use in order to create a new, (ideally) distortionfree image.