Steganography using matlab - SPIRO THE TECH GURU

www.spiroprojects.com

1.Message Analyzed: Determine message type (image or text) and dimensions (Width/Height for images; Length for text).

o Color Images are converted to be Grayscale Images

2.Header Prepared: Header data added to beginning of message for Decoding Step.

o Image Header Format: 4 digits of Width Dimension followed by 4 digits of Height Dimension

o Text Header Format: 't' followed by 7 digits of Length Dimension

3.Encryption Applied: Header and Message encrypted using symmetric XOR encryption key.

4.Message Encoded: Encrypted Header/Message encoded onto Cover Medium.

        1. Message Decoded:
           §   Header decoded/decrypted first to determine recovery cycles needed.
           §   Algorithm recovers encrypted Message data.
        2. Message Decrypted: Message decrypted using same symmetric XOR encryption key used
          earlier.
        3.Message Recovered: Message put into viewable format.
           §   Text Messages saved directly to .txt files.
           §   Image Messages require dimensions from header to be put into .bmp files.

2. Sequential vs Pseudo-Random Encoding/Decoding

Sequential Encoding/Decoding

Process:

1. Message Data is Encoded/Decoded from some starting point
(Typically upper left pixel)

2. Message Data is then Encoded/Decoded in a set unvarying pattern
(Typically to adjacent pixels)

Advantages:

· Simple to Implement

Disadvantages:

· Easily Detected using Histogram Analysis
(Can also detect Message Size Directly)

· Slow and Tedious Recovery
(Usually implemented using counters; slows recovery process)

· Repeatable Encoding Method DECREASES Effectiveness of any Encryption Techniques

Pseudo-Random Encoding/Decoding

Process:

1. Pseudo-Random Number Generator Initialized
(typically no set starting point)

2. Message Data is then Encoded/Decoded based upon the pixel location determined by Random Number Generator
(typically no set pattern)

Advantages:

· No set Encoding/Decoding Pattern for Histogram Analysis to Detect

· Quicker Recovery Rate
(Usually implemented by Pre-Defining encoding pattern; more efficient for recovery process)

· Message size difficult to estimate

Disadvantages:

· Detectable using varying sized windows and localized Histogram Analysis