17-05-2019, 07:33 AM
In the opening post, Mark wrote:
"I am completely convinced that the top left hand loop of the gallows characters is a "4" rather than a "q", but I am intrigued as to how much disagreement there is in this."
Unfortunately, he didn't let us know whether he meant medieval 4, modern 4, or any old 4 (which could include quite a bit of variation).
Also, he didn't say "shape of 4" in the first post, he said, is a "4" rather than a "q", implying that he's talking about letters.
In his second post, he wrote:
"For me this is an important question as it has implications in terms of the script used, some scripts may be more likely to have the kind of angular "4" shape in their script and other might not."
Thus, implying that he was interested in scripts. However, later, he said he was not evaluating the shape in the context of scripts.
He also wrote:
"If one can transform the problem from the context of the Voynich manuscript to just images containing circles or triangles then one transform it into an image recognition problem."
This is familiar territory for me. I'm keen on computer graphics, and I'm keen on automating this kind of process so the computer does most of the work. This comment is what prompted me to mention that an algorithm could be used to evaluate the shape and be generalized to other shapes. My comment received no response.
However, to do image recognition you need references, so you can establish the extents for the amount of variation. Each shape will have its own profile because in script (which Mark says he doesn't want to use as a reference), some variations are meaningful and some are not, and the tolerances for each character and for each style of script will be different.
Whether it's worth the time depends on how many shapes will be evaluated and the end-goal. Personally I don't think it would substantially assist in the decryption of the VMS but that doesn't mean I would discourage someone else from doing it, as they may have some other goal in mind.
"I am completely convinced that the top left hand loop of the gallows characters is a "4" rather than a "q", but I am intrigued as to how much disagreement there is in this."
Unfortunately, he didn't let us know whether he meant medieval 4, modern 4, or any old 4 (which could include quite a bit of variation).
Also, he didn't say "shape of 4" in the first post, he said, is a "4" rather than a "q", implying that he's talking about letters.
In his second post, he wrote:
"For me this is an important question as it has implications in terms of the script used, some scripts may be more likely to have the kind of angular "4" shape in their script and other might not."
Thus, implying that he was interested in scripts. However, later, he said he was not evaluating the shape in the context of scripts.
He also wrote:
"If one can transform the problem from the context of the Voynich manuscript to just images containing circles or triangles then one transform it into an image recognition problem."
This is familiar territory for me. I'm keen on computer graphics, and I'm keen on automating this kind of process so the computer does most of the work. This comment is what prompted me to mention that an algorithm could be used to evaluate the shape and be generalized to other shapes. My comment received no response.
However, to do image recognition you need references, so you can establish the extents for the amount of variation. Each shape will have its own profile because in script (which Mark says he doesn't want to use as a reference), some variations are meaningful and some are not, and the tolerances for each character and for each style of script will be different.
Whether it's worth the time depends on how many shapes will be evaluated and the end-goal. Personally I don't think it would substantially assist in the decryption of the VMS but that doesn't mean I would discourage someone else from doing it, as they may have some other goal in mind.