Jigsaw solver

[bearophileHUGS]

This can be interesting:
http://science.slashdot.org/science/.../2340238.shtml

Bearophile

[mensanator]

be************@lycos.com wrote:

This can be interesting:
http://science.slashdot.org/science/.../2340238.shtml

Bearophile

Hey, that DataGlyph demo works pretty neat.

<quote>
Decoding your DataGlyph.

DataGlyph decoded successfully.

four score and seven years ago our fathers brought
forth on this continent a new nation conceived in
liberty and dedicated to the proposition that all
men are created equal

To test the robustness and error correction DataGlyphs
provide, you might want to "damage" a DataGlyph by
editing it with paint program, obscuring some of the
marks and uploading the damaged image for decoding.

Serious aficionados have been known to print out
their DataGlyph, spill some coffee on it, tear off
small pieces, fold it into a paper airplane, run over
it with a car, light small portions on fire, then scan
the document back in and try to extract the message.
If you are such an enthusiast, have fun, be safe, and
remember even great technologies have their limits.
</quote>
Of course, being an old System Test Engineer whose job it
was to figure out how to break software, I couldn't let
this challenge go unanswered.

So, picking up the gauntlet, I broke it in 5 seconds.

<quote>
Decoding your DataGlyph.

Decoding dataglyph59683-2.bmp...
Unable to decode this dataglyph

Top 5 reasons a DataGlyph might fail to decode

1. Hey... that's not a DataGlyph
2. Not enough of the DataGlyph is present
3. Way too much damage
4. Trouble locating the DataGlyph
5. Cosmic rays

Usually if a human being can make out the individual
marks, the decoding software should also be able to
do so. This online decoder is optimized for scanned
images from a paper documents.

If you are a technical user evaluating DataGlyphs,
please be aware that this online DataGlyph decoder
is expecting mainstream DG500 DataGlyphs. This online
decoder will not work if you are testing a specialized
DataGlyph variant such as serpentone, yellow DataGlyph
or address carpet.

Finally, note that a customized image processing
front-end or a customized DataGlyph locator algorithm
might yield better results in application scenarios
such as handheld camera images with severe perspective
distortion.
</quote>

Naturally, the real answer is none of the above.

And the damage can be undone in 5 seconds also.

And, under the right circumstances, an undamaged
DataGlyph could suffer the same fate (which also implies
that the damaged DataGlyph could be read under the
same circumstances).

ObPuzzle: how did I "damage" the image?

[Tim Churches]

me********@aol.com wrote:

be************@lycos.com wrote:

This can be interesting:
http://science.slashdot.org/science/.../2340238.shtml

Bearophile
Hey, that DataGlyph demo works pretty neat.

....
Of course, being an old System Test Engineer whose job it
was to figure out how to break software, I couldn't let
this challenge go unanswered.

So, picking up the gauntlet, I broke it in 5 seconds.

....
Naturally, the real answer is none of the above.

And the damage can be undone in 5 seconds also.

And, under the right circumstances, an undamaged
DataGlyph could suffer the same fate (which also implies
that the damaged DataGlyph could be read under the
same circumstances).

ObPuzzle: how did I "damage" the image?

You created a mirror image.

The system can be made resistant to that problem by only allowing
palindromic messages to be encoded, such as "Madam I am Adam.", 'Able
was I ere I saw Elba." and "Named under a ban, a bared nude man."

Seriously, I am surprised that the Xerox demo does not try flipping the
image around various axes. It would be trivial to add these
transformations. Well, trivial to flip images along a few, obvious axes,
but not along every possible axis.

What I want to know is whether any open source implementations of this
technology are available. No doubt it is patented to death by Xerox.

Tim C

[mensanator]

Tim Churches wrote:

me********@aol.com wrote:

be************@lycos.com wrote:

This can be interesting:
http://science.slashdot.org/science/.../2340238.shtml

Bearophile
Hey, that DataGlyph demo works pretty neat.

...

Of course, being an old System Test Engineer whose job it
was to figure out how to break software, I couldn't let
this challenge go unanswered.

So, picking up the gauntlet, I broke it in 5 seconds.

...

Naturally, the real answer is none of the above.

And the damage can be undone in 5 seconds also.

And, under the right circumstances, an undamaged
DataGlyph could suffer the same fate (which also implies
that the damaged DataGlyph could be read under the
same circumstances).

ObPuzzle: how did I "damage" the image?

You created a mirror image.

Damn. Too easy. But think how much fun you could have
with a salesman if your DataGlyph was printed on a
transparency (hence, the right circumstances being
right side up or upside down).

I once made an employee id badge that I would slip
into the salesman's stack of test badges. The employee's
name was CHECKSUM ERROR. What a laugh riot.

The system can be made resistant to that problem by only allowing
palindromic messages to be encoded, such as "Madam I am Adam.", 'Able was I ere I saw Elba." and "Named under a ban, a bared nude man."
Why couldn't they simply encode your input as a palindrome?

Input: Mary had an aeroplane
Encoding: Mary has an aeroplaneenalporea na dah yraM

And then simply divide in half when decoding.

Seriously, I am surprised that the Xerox demo does not try flipping the image around various axes. It would be trivial to add these
transformations. Well, trivial to flip images along a few, obvious axes, but not along every possible axis.
That's why it took me all of 5 seconds - the time needed to
select the Flip Horizontal menu item in Windows Paint.

What I want to know is whether any open source implementations of this technology are available. No doubt it is patented to death by Xerox.
Has that ever stopped anybody?

Tim C

[Tim Churches]

me********@aol.com wrote:

Tim Churches wrote:

me********@aol.com wrote:

ObPuzzle: how did I "damage" the image?

You created a mirror image.

Damn. Too easy.

There is another explanation for the rapidity with which your brain
teaser was solved, but modesty prevents me from mentioning it.

The system can be made resistant to that problem by only allowing
palindromic messages to be encoded, such as "Madam I am Adam.", 'Able was I ere I saw Elba." and "Named under a ban, a bared nude man."

Why couldn't they simply encode your input as a palindrome?

Input: Mary had an aeroplane
Encoding: Mary has an aeroplaneenalporea na dah yraM

I should have been more precise and said "natural language palindrome",
the rules for which promulgated by Martin Gardiner of Scientific
American fame are that capitalisation, whitespace and punctuation are
ignored and that each word must be a recognised word or commonly used
proper noun in the target language (and it must read the same forwards
and backwards).

I wonder, do ideographic languages have palindromes?

No doubt it is patented to death by Xerox.

Has that ever stopped anybody?

Ask the Xerox legal department.

Tim C