posted on Nov, 30 2015 @ 09:39 PM
a reply to: IsaacKoi
All the various scans showing the text of the “Ramey memo” are generated from part of the single photograph of Ramey in which the text is
So they have a physical printout of that photograph and they are scanning it and photographing it to generate all the other images? If that's the
case, the limiting factor is really the resolution of the original image. Making more and more scans of that image is not going to help, all we need
to focus on is the original image. Assuming the digital copy captures all the detail of the physical copy then it's all we need.
Also, I did some brief reading on super-resolution processing because I thought it could only work if you had multiple images taken at slightly
different times or angles. Then you could combine them to create a single image with a higher resolution. But it seemed to me that it would be
impossible to pull any extra resolution from a single image because it's basically just pulling extra information out of thin air.
However it turns out there is actually a way to enhance the resolution of a single image using some very clever techniques. It is sort of cheating in
a way because what it does it use similarities between different parts of the same image. For example if the picture contains a bunch of flowers which
look similar, you can combine them to create a more detailed representation of the flower with a higher resolution.
In the case of text I think this technique could actually work quite well because it could combine different letters which look similar to guess what
the actual letter is. However the text on the Ramey Memo is at such a low level of detail it may be impossible to actually apply such methods to the
image. The letters themselves barely seem to cover more than a pixel on the original image so it might be a pointless effort.
I think the approach mentioned by grey580 will probably yield the best results. Knowing the number of letters in each word does actually reveal quite
a lot of information. The sequencing of words in English sentences is often extremely predictable if you know just a small number of words and their
position in the sentence. I wrote an algorithm once to predict the next word in a sentence based on the last couple of words.