Research On Image Encryption And Encrypted Image Search Based On Compressed Sensing

Image encryption has always been an important research topic in the field of informationsecurity.In recent years,with the development of new network technologies such as cloud computing,how to realize the retrieval of encrypted information has become an important topic in the field of information security ?The key problems need to be solved is how to retrieve information effectively in big mass of the image information and how to ensure the information not to be leaked in the retrieval process.In this paper,three methods are proposed.One for image encryption based on compressed sensing,and the other two for encrypted image search.Details of these methods are explained as follows.(1)Compressed sensing technology.Three aspects on the compressed sensing theory are studied,including image sparse,measurement matrices and the reconstruction algorithms.The advantages and disadvantages of different measurement matrices and reconstruction algorithms are also analyzed.(2)Image encryption and decryption based on compressed sensing.A method of image encryption and decryption based on discrete wavelet transform and generalized circulant matrix is proposed.Firstly,the Mallat fast algorithm is applied to discrete wavelet transform of encrypted image to realize image sparse process.Then,a generalized circulant matrix is used to construct the measurment matrix,in order to improve the shortcomings of the independent randomness of the cyclic matrix and obtain better reconstruction results.Thirdly,scrambling is used to eliminate statistical characteristics of the encrypted image.Finally,the security of image encryption technology based on compressed sensing is analyzed.Experimental results show the proposed image encryption technology(a)is feasible and reliable;(b)has a strong anti interference of Gauss noise and salt and pepper noise;(c)can resist shear attack when the attack ratio of the image is below 50%;(d)is strong enough to resist crack attack.(3)Methods for retrieving encrypted images.Two methods for solving the problem of encrypted image retrieval are proposed.The first one construct a retrieval unit which is embedded in the key image information through image construction of fixed format,and it can complete image retrieval without decryption.The second method adopts a non fixedformat in which keyword is embedded in the image encryption by the LSB algorithm,and it will spread encrypted keyword information into the entire ciphertext space to improve security.