Study On Reversible Data Hiding Algorithm In Encrypted Image

With the development of technology and society, digital images have been spread and utilized wider and wider. The problem of information security of digital images emerges simultaneously. Encryption can conceal the original content of images, which can prevent unauthorized users from accessing. Data hiding can embed secret information into plain image, which can ensure the security of data transmission. In a real world application, data hiding needs to be done in an encrypted image for privacy. Data management directly in encrypted domain has realistic significance and potential applications.Image owner firstly encrypts image. Data hider then embeds secret data into encrypted image without knowing any information of the original image. Receiver extracts the embedded data and recovers the image. The scheme is reversible if there is no any error in extracted data and recovered image. However, there are several problems in the existing reversible data hiding for encrypted image schemes: the first is that the reversibility which implies exact data extraction and perfect image recovery cannot be ensured; the second is the low embedding rate, which means the embedding capacity is not high; the third is that some schemes are not processed in the really encrypted domain, which requires vacating room before encryption in plain image.Based on the analysis of the existing problems, the two categories are studied to achieve reversible data hiding in encrypted image, including “do pre-operation before encryption” and “vacate room after encryption”. In this thesis, we propose high capacity data hiding reversible algorithms in encrypted images, which can guarantee the information security of digital images. The main contributions of this thesis are summarized below:(1)The security of digital images can be improved further by effectively combining image encryption with reversible data hiding scheme. By researching the existing reversible data hiding schemes in encrypted images, systematic knowledge is obtained which can inspire us to propose our own algorithm. The propose schemes can achieve real reversibility and the embedded capacity is also improved.(2)Instead of embedding data in encrypted image directly, the content owner firstly uses interpolation technique to estimate whether the location can be used for embedding and generates a location map before encryption. Next, the data hider embeds the additional data through flipping the MSBs(Most Significant Bits) of the encrypted image according to the location map. At the receiver side, before extracting the additional data and reconstructing the image, the receiver decrypts the image first. Experimental results demonstrate the proposed method can achieve real reversibility which means data extraction and image recovery are free of any error. Moreover, our scheme can embed more payloads than most existing reversible data hiding schemes in encrypted images.(3) The property of the additive homomorphism is utilized, two neighbor pixels are divided into a group, and the same encrypted key is used for each group. And then the data hider calculates the difference between two pixels in one group. All the differences generate a difference histogram. The data hider embeds data by difference histogram shifting. And the embedding capacity is improved by multi-level difference histogram shifting. The receiver extracts the additional data first, and then reconstructs the image. Experiment results demonstrate that the proposed algorithm can ensure the privacy security of image, and its embedding capacity is higher. After extracting data completely, it can recover the original image reversibly.