Imagescure

Imagescure transforms images in order to make it difficult for computers to recognize it but still recognizable by humans.

The results of Imagescure can be used to protect web against:

Two transformatins are implemented: disrupting and curving. These have been implemented with integer arithmetics only.

Disrupting is based on adding random noise to the image. This disqualificates the simplests edge detection algorithms, but should be properly configured, becase increasing the propability of disrupting corrups the readibility for humans as well.
Disrupting example 10 Disrupting example 15 Disrupting example 20 Disrupting example 25
Disrupting example 10 Disrupting example 15 Disrupting example 20 Disrupting example 25

Curving envelops the image to the region determined by random sine curves. This deteriorates pattern matching. The following example shows how the transformation looks like:
Curving example 1 Curving example 2 Curving example 3
Curving example 1 Curving example 2 Curving example 3

To further improve the protection try to combine other techniques such as increasing symbol domain (font variation), kerning that letters touch each other, etc. A good approach is to generate the password from syllabes - it is easier for humans to read it and you can apply more aggresive transformation parameters.

Happy obscuring: imagescure-0.1.tar.gz.

Viliam Holub