Reversible Data Hiding Algorithms In Encrypted Domain Based On PNG Image And Predictive Error

With the rapid development of the technologies of Internet and multimedia,there are manyimages on the Internet.Due to the convenience of cloud storage,more and more people would like to upload their images to the cloud.If these uploaded images are not protected by efficient techniques,privacy information may leak.To ensure content security of uploaded images,people often exploit image encryption techniques for protection and consequently there 'are many encrypted images in the cloud.Therefore,research of reversible data hiding for encrypted images is an important topic of information security.It.can be used in many applications,such as cloud storage,medical insurance,and military,and has important theoretical value and practical significance.This thesis investigates data hiding algorithms for encrypted images.It firstly researches several typical data hding algorithms,and then presents two new reversible data hiding algorithms in encryption domain.The specific work are as follows.1.A reversible data hiding algorithm in encrypted domain based on PNG image was proposedAccording to the characteristics of PNG image,the proposed algorithm embeds secret message into the original image and the alpha channel of PNG image,and improves embedding capacity and visual quality of the stego image.More specifically,content owner encypts his/her original image with secret key and generates an ecrypted image.Data hider creates an alpha channel with the same size of the encrypted image,and embed LSB of encrypted image pixel into the LSB of its corresponding transparent value in the alpha channel.Secret message is divided into segments containing 8 bits.For each segment,it is further divided two parts.The first part is the first seven bits,which is used to replace the 7 MSBs of the transparent value in the alpha channel.The second part is the 8th bit of the segment,which is embedded into the LSB of encrypted image pixel.Finally,an encrypted PNG image is generated by combining the stego encrypted image and the alpha channel.The receivers extract the information and recover the image separately by grasping the key number.Theoretical analysis and experimental results show that the proposed algorithm has high embedding capacity and good visual quality.2.A reversible data hiding algorithm in encrypted domain based on prediction error was ProposedThe proposed algorithm combines sparse coding with prediction error,and can improve visual quality of the setgo image,and guarantee correctness of data extraction.It firstly constructs a smooth image from the original image.Specifically,the original image is divided into non-overlapping blocks and pixels of.each block are sorted to make smooth block.For each smooth block,it is further divided into overlapping blocks.Fluctuation value of each block is calculated and the biggest one is selected as the fluctuation value of the smooth block.If the fluctuation value of the smooth block is smaller than a pre-defined threshold,this block is used to embed message.Otherwise,the block is not used in data embedding.For those blocks used for embedding,the indices(positions)of the sorted pixels in the original block are recorded.For those blocks not used,its pixels are restored to their original positions,and the indices with zero value are stored as a sign.After smooth image is constructed,content owner encrypts the smooth image,and data hider codes secret message by sparse coding and conducts secret message embedding according to the generated indices.If index values of the block are all 0,this block is not used for data embedding.Otherwise,it is used for embedding.For the used block,it is further divided into overlapped sub-blocks and each sub-block is only embedded one bit.If the embedded bit is '0',pixels of sub-block are not changed.If the embedded bit is '1',the n-th bit of the center pixel of the sub-block is altered.At the receivers'side,the receiver first decrypts the index encrypted image and segments it into a number of nonoverlapping blocks.Using the index image to check out the marked image blocks,for each of which partitions it into overlapping blocks and data extraction is then performed using the prediction error between the central pixel of block and it's four neighboring pixels.After sparse decoding the extracted data,the original embedded data is obtained.Finally,the original image can be perfectly recovered by reverse sorting marked blocks according to the index value of flag image.Experimental results show that the proposed algorithm can generate stego image with good visual quality and has no error during data extraction.