Abstract
Steganography is a theory of information hiding. It prevents secret message from being stolen and destroyed during communication. This thesis proposes a high-capacity steganographic system for data hiding. We adopt a texture synthesis algorithm based on Markov Random Field (MRF) model [Chen2003] to synthesize cover images with user-selected size. Generally, the length of a secret message to be hidden is limited by the size of the cover image; we solved this problem by using the texture synthesis algorithm. In addition, to enhance the security, we encrypt the secret message before embedding it into a cover image. The encryption method is based on the property of primitive root mentioned in Number Theory. When ? is a prime number, and ? is a primitive root of ?, {?? mod ?, for 1 ≤ x ≤ ? − 1} = {1,2, …, ? − 1}. The relationship of one-to-one correspondence could be used to rearrange characters in message and destroy the structure between them. Finally, each character is partitioned into two 4-bit words, and then embedded into two pixels of the selected cover image by LSB replacement method. The embedding capacity of an M × N image is (M × N)/2 characters.