Implementation of Least Significant Bit Image Steganography and its Steganalaysis

1
Third Quarter
Implementation of Least Significant Bit Image
Steganography and its Steganalaysis
By Deniz Oran
2
Goals

• To conceal either an image or a text message into
  a "carrier" image with Least Significant Bit
  Steganography
•  
• To detect that an image has been compromised or
  altered and to extract that message
•  
• To extend the method to apply to various image
  formats
•  
• To make the program usable with a Graphical User
  Interface (GUI)

3
What is Steganography?

• Literally means "concealed writing" in Greek
•  
• Opposite of cryptography, in which a message is
  made unintelligible, but transmitted through
  public means
•  
•  Steganography is the concealment of the fact
  that a  message is even being transferred.

4
Historical Methods

• Shaving hair
• Wax tablets
• Jargon
• Cardano grille
• Invisible Inks
• Thermo-activated inks
• Micro dots
• Enigma Machine
• All of the above have been used for espionage and
  for coordinating attacks on other bodies.

5
Current Method

• Hiding an entire text or image binary in either
  an image or audio carrier using Least
  Significant Bit encoding
• Previously done as a Senior Technology Lab using
  the same technique with .wav audio files in
  2007.
• There is currently no established method for
  encoding something with steganography because
  better encoding methods translate to lower
  carrier image capacity.
• There is no established method for universally
  detecting steganography because of the variety of
  methods.

6
Least Significant Bit Encoding

• Images on computers are made out of pixels
• Each pixel is represented by three bytes
• Each byte is composed of eight bits or 1s and 0s

10010101 00001101 11001001 10010110 00001111
11001011 10011111 00010000 10010100 00001101
11001000 10010110 00001110 11001011 10011111
00010001

• The second set of bytes represents the encoded
• ASCII character G (01000111)

7
Program Design in JAVA

• Create a GUI
• Dropdown menu
• Input a suspected image
• Attempt to detect if it has been compromised
• Place the two image next to each other for
  visual comparison when extraction is finished.

8
Inputing the Image

• Read the binary of the image using Buffered
  Reader, DataBufferByte, and WritableRaster into a
  three dimensional array (data cube) with the row,
  column and corresponding color value (depends on
  image format).
• Alter every Least Significant Bit in each byte in
  order to spell out the word. The ASCII values
  are understood by DataBufferByte instead of
  having me define each one.
• Save the altered image because it is ready to be
  sent
• Saving should be done using a lossless image
  standard because this form of steganography
  exploits the noise of an image. Therefore,
  .PNG is a good start.

9
Steganalysis

• Detecting a message encoded using steganography
  is only feasible if the original image is
  available or if the exact encoding method is
  known.
• The program will still be able to detect if an
  image has been compromised by converting the
  suspected image into hexadecimal but the message
  may not be extracted.
• This method wont work if a proper LSB is done,
  since the encoding starts after the data for the
  hex tags, thus not altering them.

10
Decoding

• Just attempt to reverse the process of Least
  Significant Bit encoding. This method is always
  tried by commercial software but hardly works if
  the LSB is done differently.
• The suspected image is inputted, converted to
  binary, the length of the message was hidden in
  the first 32 bytes, that length is the parameter
  for the for-loop which traverses the rest of the
  image extracting the least significant bits in
  the same manner they were isolated when they were
  inserted
• The output is the original ASCII message
  inputted by the first user

11
Possible Problems

• Steganography is an increasingly complex field
  because of new algorithms and unique ways of
  analyzing images
• There are many methods to inconspicuously hide an
  image or text message with computer science, some
  are less complex (microdots) and some use higher
  level mathematical algorithms (stochastic
  modulation)
• The Least Significant Bit method is an
  established method and can be detected at 80
  accuracy using highly advanced math but not the
  human eye

12
Testing

• Tests involve inserting a character into an
  image, finding where the character is hidden, and
  verifying that it is indeed the letter. Tests on
  file size versus message length as well should be
  done. Visual comparison is required.

Image 1 Image 2
13
Testing continued
Binary comparison revealed that after 32 bytes
there was a variation when the letter G
(01000111) was encoded Image 1 Image 2 There
was also testing on the rendering time of the
image, the time it took to encode and decode the
image with varied message lengths and image
sizes. The trend appeared linear for each
process, but the slopes of the curved varied.
14
Learned Concepts

• How images of various formats are constructed,
  especially .PNG.
• How to manipulate binary code and understanding
  the role of bits within bytes. Binary operations
  such as AND, OR, ADD.
• Various methods of Steganography and
  Steganalysis and which are the most effective
• How to determine a file's extension using JAVA

15
Future Plans

• Extend the program to work with other image
  formats and improve the algorithms susceptibility
  to common detection methods (using padding and
  random byte locations)
• Pairing the steganography with a simple
  encryption algorithm