Research On Ciphertext Information Hiding Algorithm Based On Block Strategy

Information hiding in ciphertext domain is an important research field in the field of cryptography.It is widely used in digital signature,copyright protection,tamper detection,traitor tracking and etc.Encryption is an effective technology to protect the content of the host image,while information hiding can achieve the purpose of information transmission by embedding additional data into the host image.Based on their advantages,embedding additional data in the ciphertext domain can realize the transmission of secret data under the premise of protecting image privacy.Image itself has the inherent characteristics of large capacity,high redundancy,strong correlation,etc,and the existing encryption technology aims to destroy these inherent characteristics to achieve the purpose of privacy protection,like pixel-level image encryption technology and block-level image encryption technology.The former almost completely destroys the correlation between pixels,making it difficult to encrypt additional data in an encrypted image,and the latter keeps the intra-block correlation on the basis of the block-splitting strategy and is widely favored in the field of information hiding in ciphertext domain.In this paper,based on the block strategy,according to the functions of the embedded additional data,we carried out our research respectively from the digital fingerprint for traitor tracking and information hiding for secret communication.In the following,we describe the main works in this paper.(1)On the basis of block strategy and vector quantization,a joint fingerprint decryption scheme based on the weight function is proposed.First,the original image content is segmented and the vector quantization process is completed on the segmented image content.The scrambled and replaced codeword indices are transmitted to the receiver.The receiver divides the codeword indices into a series of blocks,and calculate the weight value of each block's codeword index through weight function;by modifying only one codeword index,the n+1 bit fingerprint sub-sequence is embedded into one block containing 2~n codeword indexes,which ensures that the fingerprint sub-sequence is embedded while decrypting.In this scheme,the bandwidth and cost consumed in the transmission process are effectively reduced by the sender,and the blind fingerprint extraction is realized while the flexible fingerprint embedding is guaranteed by the receiver.(2)Based on the division strategy and the homomorphic encryption,a high-capacity reversible data hiding in encrypted images using cross-channel correlations is proposed.In the encryption stage,the content owner first obtains the key matrix generated by the 3d Logistic mapping,encrypts the R,G and B channels of the color image with the obtained keys and sends the encrypted content to the receiver,and homomorphic modulo addition is adopted in the encryption process.In the data hiding stage,the receiver first generates the cross-channel correlation between R,G and B channels,calculates the difference histogram between the adjacent pixels of each cross block,and performs the embedding of additional secret information through the histogram shifting.In the stage of data extraction and image recovery,the receiver completes the correct extraction of additional information and the perfect recovery of the host image separately according to the encryption key and the information hiding key,that is,the information extraction operation can be completed before or after the image decryption.Analysis and experiments demonstrate that the proposed homomorphic encryption is secure,and the performance of RDH is superior.Under the same PSNR the embedding rate in each channel of the encrypted color image with our proposed method is larger three times than that shown in the state-of-the-art publications;by using histogram shifting,this algorithm can correctly extract additional information and perfectly recover the host image content.