DCT Stenography
- Thread starter aaronb50
- Start date
- Local time
- Today, 03:27
- Joined
- Jan 23, 2006
- Messages
- 15,555
You may also find this link of value when looking into BMP formatting.
Also the OR is for bit setting
eg
?5 or 3
7
where 5 is 101 and 3 is 11 respectively in binary format
If you OR 101 and 11 you get 111 which is 7
Also, the Chr(asc()... gets you the character for the ASC value
?chr(ASC("x"))
x
? chr(asc(5) and 254)
4
and
? chr(asc(5) and 254) OR 3
7
For more info on MS ACCESS Boolean Operators (more than you probably want)
From an earlier link
More steganography details here: and here
Also the OR is for bit setting
eg
?5 or 3
7
where 5 is 101 and 3 is 11 respectively in binary format
If you OR 101 and 11 you get 111 which is 7
Also, the Chr(asc()... gets you the character for the ASC value
?chr(ASC("x"))
x
? chr(asc(5) and 254)
4
and
? chr(asc(5) and 254) OR 3
7
For more info on MS ACCESS Boolean Operators (more than you probably want)
From an earlier link
More steganography details here: and here
Last edited:
- Local time
- Today, 08:27
- Joined
- Sep 12, 2006
- Messages
- 16,024
I assumed that a jpeg image (or even a bmp image) consists of something like this
maybe a bit length (1-255) followed by a 24bit colour value - so the image consists sequences of bits of a given colour, although I can see that this would tend to a massive image size for images consisting of no image at all - just random pixels.
Anyway - assuming a jpeg holds something of this nature, then you could adjust the colours by xoring the colour bits with some "known" value, which, as I say, would have the sole effect of changing the colour of each pixel block.
Now to recover the original image, you would repeat the process, since a xor b xor b = a again. So if someone had the original jpeg, they can recover the changes (and hence the message) by doing a similar bit wise process comparing the new image with the original.
And an observer would just see a normal jpeg, with strange colours.
That's what would make sense to me. Now does it make sense with this algorithm?
----
now I've read the linked article above, and it is something like this, I think.
changing the least significant bit of any colour (the R, G or B component) gives a tiny variation of colour, but hardly detectable visually, (so not strange, as I had assumed) although apparently, if you know it is there then maybe relatively easily detected by analysis as there is no encryption
maybe a bit length (1-255) followed by a 24bit colour value - so the image consists sequences of bits of a given colour, although I can see that this would tend to a massive image size for images consisting of no image at all - just random pixels.
Anyway - assuming a jpeg holds something of this nature, then you could adjust the colours by xoring the colour bits with some "known" value, which, as I say, would have the sole effect of changing the colour of each pixel block.
Now to recover the original image, you would repeat the process, since a xor b xor b = a again. So if someone had the original jpeg, they can recover the changes (and hence the message) by doing a similar bit wise process comparing the new image with the original.
And an observer would just see a normal jpeg, with strange colours.
That's what would make sense to me. Now does it make sense with this algorithm?
----
now I've read the linked article above, and it is something like this, I think.
changing the least significant bit of any colour (the R, G or B component) gives a tiny variation of colour, but hardly detectable visually, (so not strange, as I had assumed) although apparently, if you know it is there then maybe relatively easily detected by analysis as there is no encryption
Last edited:
