| In recent decades,the digital steganography technology has been widely used in digital authentication,copyright protection and multimedia digital products.The traditional digital steganography algorithm embeds the secret message at the expense of modifying the original carrier data,this change is irreversible.To solve this problem,the reversible steganography is proposed,in which the original carrier image can be restored after extracting the hidden information without distorition.It usually be applied to the field of medicine,military,legal evidence image and others.The reversible steganography has made great progress,however,it still need to be further studied in the aspects of embedding capacity and robustness.Firstly,this thesis introduces the basic algorithms of image steganography and reversible steganography.Then,two improved algorithms are proposed which based on reversible steganography algorithm with general difference expansion and reversible steganography with IWT and general difference expansion:(1)A generalized difference expansion reversible steganography algorithm based on image block is proposed.In the existing generalized difference extended reversible steganography algorithm,it ingores the correlation between piexl blocks and the influence of embedding direction to image quality and embedding capacity.In order to solve this problem,a new reversible steganography algorithm is designed by combining the judgment method of block type and the block direction,it can select the smoothing block and the appropriate direction to embed information,while do take into account the block type and the embedding direction.Experiments show that this algorithm can both improve the steganography capacity and gurantee perceptual quality.(2)A generalized difference expansion reversible steganography algorithm using the bit plane theory.The problem of the large difference between the pixels in traditional generalized difference reversible steganography algorithm decrease in the number of embeddable pixel pairs,as a result,it affects information embedding capacity.In order to solve this problem,this algorithm reduces the pixel difference by combining the bit plane decomposition method,it increases the number of available pixel pairs and further improves the embedding capacity.By embedding message on thelower bit planes,the higher bits are not processed.Experiments show that this algorithm can both improve the steganography capacity and gurantee perceptual quality. |