Abstract
We have proposed a novel and secure steganographic method for embedding secret data into a gray scale image in this thesis. In the embedding process, the image is partitioned into non-overlapping blocks. The pixels in a block are classified into unchangeable pixels and changeable pixels. The average of the gray values of unchangeable pixels in each block is calculated as . The embedded amount of data for each changeable pixel is determined by the difference between and the gray value of the changeable pixel. A larger difference embeds a higher amount of secret data. We modify the gray value of a changeable pixel and reset a new gray value with secret data embedded. But, not all the changeable pixels are appropriated for embedding secret data. The gray values of some changeable pixels may cause wrong extraction if they were used in embedding. We have two fault potential checking processes which execute both embedding procedure and extracting procedure to automatically detect and discard fault potential changeable pixels. Therefore, no extra space is needed for recording the positions of the fault potential changeable pixels. The experimental results show that we can embed secret data into an image without noticeable distortion.