| With the increasing popularity of cloud-storage services,the local space has been largely saved.Unfortunately,the outsourcing of data storage to the cloud raises great security and privacy concerns.To provide privacy guarantees for sensitive data,a s-traightforward approach is to encrypt the sensitive multimedia data locally before out-sourcing.Once receiving the encrypted data,the cloud server can embed some water-mark information such as user's ID and authentication information etc to enable query-ing,managing and identification.The cloud server has no right to introduce permanent distortion into user's data.It should be clear to perfectly recover the original data after the watermark is extracted correctly.Reversible data hiding technique for encrypted images is a special watermarking technique with the characteristic of not only the message can be precisely extracted but also the cover itself will be restored and encrypted in a lossless way.The watermark embedded in the image can achieve the goal of both the integrity and authentication.The proposed algorithms in the past just take invisible watermark embedded by the cloud server into consideration.The reversible data hiding technique adopted by the cloud server is relevant to either the encryption methods by the sender or the decryption methods by the receiver.However this can not be applied in reality.What is worse,it does not consider these following application scenarios.(1)Since the cloud is thought to be only semi-honest,the client also needs to embed watermark to ensure the integrity and authentication.(2)The cloud server may embed visible watermark to easily recognize the owner of encrypted data.Based on all the application scenarios mentioned above,this dissertation studies different reversible data hiding techniques for encrypted images.it can be summarized as the following three aspects:1.Reversible data hiding method by client based on unified embedding and scramblingTo enable the client to embed invisible watermark,the dissertation proposed an unified embedding and scrambling technique called UES which can embed a large amount of message into the image.After doing this,the goal of encryption can be achieved.In this dissertation a novel pixel intensity value method called local linear prediction is proposed to accurately predict the pixels in the image.Thus large payload can be obtained.2.Reversible data hiding method in encrypted images by cloud based on re-versible image transformationThe dissertation proposed a novel algorithm for reversible data hiding in encrypt-ed images based on reversible image transformation.This method allows the user to transform the content of secret image into the content of another target image with the same size.The transformed image,that looks like the target image,is used as the "encrypted image",and is outsourced to the cloud.Therefore,the cloud server can easily embed data into the "encrypted image" by any re-versible data hiding method for plaintext images.And thus a client-free scheme for reversible data hiding method in encrypted images can be realized,that is,the data embedding process executed by the cloud server is irrelevant with the processes of both encryption and decryption.3.Reversible visible watermark method in encrypted image by cloudThe dissertation proposed a novel algorithm for reversible visible watermarking.Among all the visible watermarking methods,the plaintext image is chosen to be the cover.To achieve the reversible goal the difference between the original image and watermarked image should be recorded.Since there is no correlation in the encrypted image,it is hardly to embed a large amount of differences overhead in a reversible way.In the proposed method,we empty out room by embedding some significant bits of pixels of watermark region into non-watermarked pixels with a traditional reversible data hiding method and then encrypt the image.So the reserved positions of these significant bits in the encrypted image can be used to embed the visible watermark. |
PDF Full Text Request