Demonstration of Visual Encryption
This is an attempt to demonstrate
visual encryption
using ordinary web graphics. If you see only a field of random mottle and this text, its likely the hspace and vspace tag parameters I used are not right for your browser. You can try adjusting them but without an aid, its a tough, frustrating trial and error process.
If you see just a patch of mottle, you have your browser's background option turned off.
Without background, the gif image that produces the signature is the image on the right(a patch of mottle.) It is the overlay of this image over the background (which you can see above) which results in the rendering of my signature.
Theory of Operation
This encryption is the analog of spread spectrum modulation: two dimensions instead of time and optical instead of communication signal transmission. The source graphic is used to modulate the random background, completely in phase for white, completely opposite phase for black. Recovery depends on matching the alignment (now in two dimensions) such that the out of phase signals completely interfere and the in phase signals do not.
This technique not only works with image overlays on the computer, it also works with transparencies (eg overhead projection slides) such that no computation is required.
Construction process
Start with a random field, R1
Generate a black and white bit map, M1, with only blacks and whites (no middle tones)
Using any of several bit map editors which provide for adding two bit maps, add R1 and M1 creating I1.
Reverse both R1 and M1 (create visual negatives) and add these two bit maps creating I2.
Reverse both I1 and I2 (again creating visual negatives) and add these two bit maps creating the encrypted bit map E1.
Check by adding E1 and R1 (R1 in original polarity, positive not negative) A modulated version of M1 should be recovered.
At this point, generation of R1 is a cumbersome process. Sufficient examples for experimentation can be obtained from
http://www.cl.cam.ac.uk/~fms27/vck/share1.gif
I am thinking about writing one in postscript. All the pieces are in hand, its only a matter of making the time.
Background
A process similar to this is described by Naor and Shamir in their paper Visual Cryptography. In the case they describe, there is a different theoretical justification based on XORing using a four pixel cell, such that the XOR effect results from a simple OR of an overlay. It was during experimentation with their process that I realized that treating the process as an analog of spread spectrum modulation results in a simpler and more direct method. Further, generation of the noise mask (the one time pad) becomes much simpler, with its only constraint being that image reversal (black/white instead of white/black) must not bias the image density
John Bailey May 29, 2000
Home