Image processing, highlighting hidden details

[ReneZ]

Following are some processed images of the Voynich MS that did not involve photoshop.
They also do not try to simulate the effect of wetting and rubbing the pages. In fact, I am unsure in several cases what it is that they are showing.

They are intended to be false-colour images of the parchment.
They were made using the following procedure.

Largely speaking, in colour scale terms, the parchment has a certain hue, with a varying lightness.
One can establish the R-G-B coordinates of the lightest parchment, and set up a new colour coordinates system with one axis pointing from (0,0,0) to this point and the other two axes perpendicular to this axis.
One can then compute the distance of each pixel from the 'parchment axis' as the root-sum-square of these two perpendicular components.
That distance is then colour-coded. If it is zero, we render it black, and increasing values become increasingly colourful. If it is above a certain limit, we assume that we are no longer looking at parchment, and we use grey for the colour code.

I will show links to the images of the biological section in the next post.
In some cases we can see parts of the other side of the page (just like we can see in the normal scans).
In some cases, we can see things that are on the opposing page.
Also, we can see shading from the illumination and blemishes and folds in the parchment.

Then, in some cases there seem to be things that cannot be easily explained by any of the above.

[ReneZ]

As my web site host is not responding right now, this may take a while.
Let me just try to add a few examples here...

You are not allowed to view links. Register or Login to view.
You are not allowed to view links. Register or Login to view. 
You are not allowed to view links. Register or Login to view. 
You are not allowed to view links. Register or Login to view. 
You are not allowed to view links. Register or Login to view. 
You are not allowed to view links. Register or Login to view. 
You are not allowed to view links. Register or Login to view. 
You are not allowed to view links. Register or Login to view. 
You are not allowed to view links. Register or Login to view. 
You are not allowed to view links. Register or Login to view.

[nablator]

There are many color-space manipulation tools, one of the best I've seen is the plugin for ImageJ called DSTRECH, designed to reveal faint rock paintings. When stretching the color space considerably, only the best quality of digital images should be used, otherwise compression artifacts will be plainly visible.

[Moonchild]

Can you tell us what are the areas or an example of an area that cannot easily be explained please from the images provided above

[ReneZ]

When stretching the color space considerably, only the best quality of digital images should be used, otherwise compression artifacts will be plainly visible.

These images were made from the 400 dpi scans of 2014, but I smoothed them before.
Also I used a mask (4x4 pixels) to blank out 'known' text and drawings.
All that was done with home-brewn algorithms.

I will add the other folios as soon as the web host is behaving normally again.

Can you tell us what are the areas or an example of an area that cannot easily be explained please from the images provided above

I have not even looked at everything in detail. I would invite everyone to try, but it works best with a large screen.

[RobGea]

Decorrelation stretch algorithm in Python [ github: lbrabec / decorrstretch ]
Link: You are not allowed to view links. Register or Login to view.
Dependencies : PIL / OpenCV + Numpy

It does a nice job pulling out the colors.

[ReneZ]

Here are You are not allowed to view links. Register or Login to view..

[ReneZ]

Looking at, for example, f75r, one can see parts of the image of f75v, but below the pool there seem to be structures that do not match the other side. We do not know what used to be on the page before f75.

[nablator]

For comparison, two settings of DSTRETCH on the bottom right corner of f75r:

   

   

I doubt there is any "structure", maybe only old dirt and a more recent abrasion?

If there is an ink transfer from the opposite page there, it's hard to make out.

   

Local contrast enhanced:

   

It's not as good as my spray plaster wall for induced pareidolia.

[pjburkshire]

Once the world realizes what f75r really is, it's going to be banned as pornography.