Image-based Steganography

Hoping someone might be able to help me. How do you combat Image-based Steganography?

What is meant by "combat?"

Are you interested in detection of images that carry steganographic messages, or something else?

Write a rediculously intricate algorythim to search for all known and hypothetical digital stenography concepts? The answer, unless you have millions to spend, is: on a case-by-case basis.

I can't help thinking that you have asked this question for the sake of creating a thread. Why do you need help with this?

Interesting question, though we're not sure what exactly you are trying to combat.

One common steganography technique using an image is to use the low bit in the pixel data to carry the hidden message. The resulting image when viewed will probably appear entirely normal.

If you want to detect whether there's a hidden message, you would need to analyze the low bits from all the pixels and see whether there's a pattern. If the message is encrypted by a standard algorithm like AES the low bits will appear entirely random. If the image contained homogeneous areas (like a drawing, say) you might expect the low bits to be all the same and the fact that they were random would suggest the presence of a message.

If you have control of the image and want to prevent transmission of a hidden message, set all the low bits to 1 or 0 to erase the message. The image will probably be indistinguishable from the original.

Of course this covers only one technique.

By "combat," I'm not sure what you mean. If you mean corrupt images such that you destroy encoded information, then just tweak the image somewhat, or do what Skeptic said and set all the LSBs to zero or one. If you mean detect it, then it's going to require a lot of research on your part.

That said, I was working on a steg project (just for fun) where I hid data in images and I did some research so that I could do it in a slightly less amateurish manner. I found two (seemingly) outstanding documents that I still... haven't quite had a chance to read in their entirety. Perhaps you should take a look at them.

(It seems that new accounts are forbidden from posting URLs. Please PM me and I'll send you the links.)

I think that I do remember from one of these documents that LSB encoding tends to produce a larger quantity of unique colors in an image compared to the number of unique colors you would expect to find. That said... determining how many unique colors you should expect to find sounds... impossible (but I'm not an expert).

Good luck.