الفهرس 
Principles of SteganographyOnly 14 pages are availabe for public view
 |  | from | 186 |  |  |
| | | from | 186 | | |
Abstract
Humans are continuously attempting to find new and efficient ways to communicate secretly. A common solution to this problem is to use encryption in order to obscure the information contents of the message. However, we can think of steganography as a complementary solution. Steganographic techniques hide one piece of information inside another cover data so that they appear as a single entity.
The most simple and straightforward steganography scheme is the Least Significant Bit Substitution)LSB(, where the message bit-stream is embedded into the least significant bits of the image pixels. In this thesis, we will introduce some wavelet-based techniques for modern steganography in digital imagery. The proposed techniques promote security, maximization of payload, high imperceptibility, as well as message error free recovery.
Actually, we have implemented four algorithms for embedding a message into the wavelet transform of true-colored images. Each one of them introduces a different idea and hence provides a certain advantage over the others. The first algorithm (WLTCodedBinHide) applies the standard technique of LSB on wavelet coefficients of the cover image. It also proposes using linear block codes in order to detect and possibly correct the truncation errors that may result from the floating point representation of the wavelet coefficients.
The second scheme (IntWLTBinHide) exploits the idea of reversible wavelets to store message bits directly in the LSBs of the integer coefficients. However, this requires applying a preprocessing step on the cover image to adjust the saturated pixel components in order to guarantee that the embedded message will be recovered correctly The next algorithm (WLTCastBinHide) uses the idea of casting message bits onto normalized cover coefficient using an embedding strength parameter. The extraction process, on the other hand, involves retrieving the message bit-stream by comparing the DWT coefficients of original image with the corresponding coefficients of the stego-image to decide upon the value of the embedded bit.
The last algorithm (WLTFusedHide) is based on merging the wavelet decomposition of the normalized cover image and the normalized secret image into a single fused result using an embedding strength factor (α). The algorithm also applies an adjustment operation on the normalized cover pixels before the embedding process takes place in order to guarantee that the embedded coefficients would not go out of range and hence the message will be recovered with acceptable accuracy even with a small value of α.
Experimental results showed the IntWLTBinHide provided approximately the same capacity as the most recent spatial domain (LSB-based) hiding schemes. However, it achieved a better invisibility performance. In addition, the WLTFusedHide achieved the best invisibility performance over all the presented algorithms, while it can embed a gray-scale image that is three times the cover image size. Generally, we can say that the proposed algorithms achieved our primary goals by providing high payload, high invisibility, and good signal recovery.