luca-m / lsb-toolkit

This is a small and simple toolkit that might be useful during steganalysis, it is currently composed by several general purpose command line tools.

lsb-toolkit

This is a small and simple toolkit that might be useful during steganalysis, it is currently composed by several general purpose command line tools.

• Version: 0.3
• Author: [email protected]
• License : Attribution-NonCommercial-ShareAlike 3.0 Unported (CC BY-NC-SA 3.0)

STEGANALYSIS TOOLS

---

Install: pip2 install -r requirements.txt

lsb-enancher

Produce a picture from input image that enhance bit variations. 
This particular picture might reveal suspicious patterns inside the analyzed image.

lsb-extract

Dump the 'LSBs' of the image pixel with several modalities. For instance is
possible to select wich channel consider during extraction ([a][r][g][b]) or 
which bit consiter ([0..7]). 

Supported algorithms:
RAW LSB - examples (pixel/data-bits)
	channels=argb 
		pixel | rgba | rgba | rgba | rgba | 
		data  | 7654   3210 | 7654   3210 | 
	channels=rgb
		pixel | rgb | rgb | rgb | rgb | rgb | rgb | rgb | rgb | rgb | rgb | rgb |
		data  | 765   432   107 | 654   321   076 | 543   210   765 | 432   107 |
	channels=rg
		pixel | rg | rg | rg | rg | rg  
		data  | 76 Basic Profile
	channels=r
		pixel | r | r | r | r 
		data  | 7   6   5   4 
CLASSIC LSB - examples (pixel/data-bits)
	channel= rgb (mandatory)
		| rgb | rgb | rgb | rgb | rgb | rgb | rgb | rgb | rgb | 
		| 765   432   10- | 765   432   10- | 765   432   10E |
		 '-' means unused, 'E' means end of bit stream (our "EOF")
		 In practice '-' is a 0-bit and 'E' becomes 1-bit

lsb-embed

lsb-extract's dual tool.
Embed data from STDINPUT into a picture using the same algorithms

lsb-randomness-b

Calculate the randomness of a bit string.

lsb-autocorr-b

Calculate theautocorrelation of a bit string measuring how much the shifted
version of the bit string is similar to the original bit string.
Number of shits and shift step are configurable.

lsb-chisquare

Perform a chi-square test on an image pixel stream. This statistical test
helps to determinate when the distribution of lsb is similar to a random 
distribution (might reveal encrypted or compressed payload).
If you need graphics use some external tool like gnuplot.

lsb-echo-pixels

Utiliy that produce a pixel stream in output, in other world it 
print out the integer values of the image pixels.
Typical usage: /lsb-echo-pixels -f image.png | ./lsb-chisquare

lsb-image-create

Create pictures from a pixel stream. It could be useful for visualizing raw files 
in order to observe the presence of regular patterns, or also for recreate image 
from an extracted pixel stream (eg. with lsb-echo-pixels)

lsb-imageop

Utility for performing pixel per pixel operation on a couple of images.
Can be really useful if user has access to the original steganography carrier. 

USAGE:

---

./lsb-extract.py -f image.png -o dump -c rgb -b 012

Means that you will extract the bit 0,1 and 2 from every pixel considering rgb channels. Note that rgb is different than grb and 012 is different than 102 !

Use --help on each tool for further detail.

FUTURE DEVELOPMENT:

---

- Support audio too..
- Support more algorithms
- Support more analysis (RS analysis?)

lsb-toolkit's Issues

Does not work with Pillow

Pillow, the friendly PIL, is a PIL fork that is being maintained as opposed to PIL itself.

To support it, all occurences of import Image need to be changed to from PIL import Image

Grayscale image support

Avoid to convert to RGB grayscale images.

KeyError 'stego_algs' in lsb-extract.py

Following the README.md, when executing:

$ ./lsb-extract.py -f image.png -p dump -c rgb -b 012

I get this errror:

Traceback (most recent call last):
  File "./lsb-extract.py", line 43, in <module>
    algs=[k for (k,v) in sys.modules['stego_algs'].__dict__.items() if inspect.isclass(v) and k != 'Algorithm']
KeyError: 'stego_algs'

Refactoring

Refactor all the tools in a more elegant python project