Reversible Data Hiding In Xor-permutation Encrypted Image Based On Neighborhood Prediction

With the developing of cloud computing,digital images stored in the cloud in the form of ciphertext becomes an effiective way to protect user privacy.Meanwhile,cloud service provider needs to embed related information in ciphertext images to realize ciphertext images management.As a technique that takes account of users’ privacy protection demand and cloud service providers’ ciphertext image management demand,reversible data hiding in encrypted image becomes a research hot spot.Reversible data hiding in encrypted image algorithms based on symmetric encryption system have the advantages of low computational complexity and no data expansion,which attracts the widespread attention of researchers.In order to improve the security of symmetric-encrypted images,a simple and general image encryption scheme called XOR-permutation is designed in this paper,which is combined with bitwise exclusive-or encryption and scrambling encryption.Based on the XOR-permutation encryption scheme,to improve the quality of decrypted image as well as the embedding rate of data hiding,two secure separable reversible data hiding in encrypted image methods are proposed.The main contents of this paper are as follows:1)The statistical characteristics of encrypted images generated by our XOR-permutation scheme,XOR scheme and the XOR-block scrambling scheme are first analysed,which contains the histogram characteristics,difference histogram characteristics and the inter-pixel redundancy.Further more,the performance under the known plaintext attack is discussed.The time complexity of three encryption schemes is compared at last.Experimental results demonstrate that the encrypted images obtained by the XOR-block scrambling scheme preserved parts of the statistical characteristics of original images,while the encrypted images obtained by XOR scheme failed with the known plaintext attack.Our XOR-permutation scheme overcomes the above problems with a low time complexity cost.2)Based on the XOR-permutation scheme,a secure neighborhood-prediction based reversible data hiding method in encrypted image(RDH-EI)method is proposed.After XOR-permutation encryption,some encrypted pixels are pseudo-randomly chosen for data hiding,and secret information are embedded into the most significant bit(MSB)of chosen pixels by the bit replacement operation.In the image decryption phase,the possible marked pixels are predicted and corrected by comparing the difference between each pixel and its neighborhood average value to improve the quality of decrypted image.In the image recovery phase,for each marked pixel obtained by the data hiding key,five neighborhood templates are designed to compute its fluctuation value,which is used to deduce whether the MSB of it is changed or not.This paper discusses and analyzes the threshold selection and the prediction accuracy.Experimental results demonstrate that proposed neighborhood prediction method can correctly predict at least 96% marked pixels,and the decrypted images and the recovered iamges are of high quality,and the embdedding rate is high.2)In order to achieve complete reversible,and improve the embedding rate of data hiding,an MSB-flipping prediction based RDH-EI method is proposed.The XOR-permuation scheme is still used in proposed method.Before image encryption,original pixels are divided into four types to reserve the MSB plane according to the MSB-flipping prediction scheme.In order to provide the basis for lossless decryption,pixel types are recorded by a pixel classification matrix that of the same size as the original image.Since the four pixels can be distinguished by only 1 bit,the pixel classification matrix is a two-value matrix.The pixel classification matrix is compressed and encrypted,and then embedded in the XOR-permutation encrypten image to generate the final encrypted image.In data hiding phase,bit replacement operation is conducted to embded the secret data into the MSB of the encrypted pixels.At the receiver side,marked-decrypted image of high visual quality and decrypted image,which is the same as the original one,can be obtained only with the encryption key;with the data hiding key,secret data can be extracted form the marked-decrypted image and the marked-encrypted image,respectively.Experimental results show that proposed MSB-flipping prediction scehme makes full use of the correlation of the adjacent pixels,and the pixel classification matrix is sparse to compress,which leads to a large space for data hiding.In summerry,propsed method achieves completely reversible,and high embedding rate,more over,it is safe and flexible.