Adaptive Data Hiding In Wavelet Domain Based On Texture Analysis Of Image

The rapid development of Internet has basically changed the life style of people, and the communion and communication between people become more convenient and faster. But at the same time, it also becomes possible that the opening of Internet has made the data been attacked during transmitting. Therefore, the security of the data on Internet has become a significant title in the information security field. As a new branch of information security, Information Hiding emerges as the times required. It embeds secret information into digital media to implement covert communication.The thesis introduces the basic theory of information hiding, and studies on the hiding method based on the human visual system, and proposes an adaptive data hiding in wavelet domain based on texture analysis of image. The wavelet transform is consistent with the HVS characteristics and the second generation integer to integer wavelet transform in lifting scheme can reach the requirement of lossless image reconstruction, so 9/7 integer lifting wavelet transform is adopted and the adaptive information hiding is implemented in transform domain.The low-frequency sub-band of one-scale DWT retains a majority of energies and visual characteristics of original image; moreover the low-frequency coefficients are more robust than middle-frequency and high-frequency coefficients, so the hiding method is implemented on this low-frequency region. To avoid the image quality decline due to changing the low-frequency coefficients with traditional method, the texture analysis is proposed. The region is signed as smooth, edge and texture using entropy and variance degree, then different embedding deepness is decided by JND(just noticed difference)in different regions. It increases the hiding capacity on the premise of ensuring the image quality. In addition, resetting low bits in low-frequency sub-band of wavelet domain as pretreatment makes the blind detection easily accomplished, which increases the communication efficiency and the secret information security; the encryption of secret image based on chaotic mapping is used to further enhance the security; and the histogram of cover image is modified to avoid secret information loss due to the data overflow.According to this data hiding method, the embedding rate of complex texture image can be up to 9.13%, and the routine robust experimental results show this method is feasible and satisfactory.