Research On Image Encryption And Data Hiding In Encrypted Images Algorithm Based On Compressive Sensing

With the continuous advancement of science and technology as well as the further research of big data,people use more and more frequently the Internet to transmit information.In a fast-paced era,people prefer to reading information quickly.The vivid and intuitive characteristics of the digital images meet people's needs for information.The digital images are popular with the majority of the Internet users.Consequently,the security problems of digital images are becoming increasingly prominent during transmission.Unfortunately,some defects in the open Internet protocol are utilized to intercept the information transmitted on the Internet by the illegal users resulting in the leakage of information.Therefore,encrypting the digital images in transmission is an important way to avoid the information leakage.To improve the utilization of the resources,it is necessary to compress and encrypt images before transmission and storage with the limited storage space and bandwidth.Compressive sensing(CS),as a new signal acquisition theory,can complete compression while sampling,which effectively reduce storage space and transmission bandwidth.The chaotic systems have excellent properties such as sensitivity,pseudo-randomness and determinacy for the initial conditions and control parameters.The sequences generated by the chaotic maps not only can change the pixel positions or values of the image,but also can construct the measurement matrix.Therefore,the chaotic systems are widely used in the field of the image encryption.Based on these theories,the image encryption and the data hiding in the encrypted image algorithms are proposed in the paper.The main work and research contents are listed as follows:1.A double color image encryption scheme via CS and three improved low-dimensional chaotic systems is proposed.Firstly,the correlation coefficients of the plain image,the chaotic maps and the SHA-256 function are used jointly to generate the key streams and the measurement matrices.Secondly,the red,green,and blue components are extracted from the color images,and the DWT fixed dictionary is used to sparsely represent each component.Then,the measurements are obtained after the measurement matrices compress the image.Finally,the bit-level confusion-diffusion process and the plane-level pixel cross-confusion process are implemented to obtain the scrambled image on the measurements merged.To further enhance the security level of the cryptographic system,the key streams are employed to perform the diffusion operation on the scrambled image.The simulation results show that the cipher image is similar to noise and display nothing information about the plain image,and the decrypted images have good recovery effect.Security analysis proves that our scheme has strong resistance against attacks and robustness.2.A novel color image compression encryption scheme using the CML based on LSS map and the learning dictionary is introduced.Firstly,combining the relevant information of original image and initial secret keys,the initial values and parameters of CML chaotic system are updated by SHA-384 function.The CML system is iterated to gain three measurement matrices and the pseudo-random key streams of encryption.Then,the red,green and blue component matrices are split from the original color image,and each component matrix is sparsely represented by KSVD learning dictionary.Subsequently,the measurement matrices of three components are used to measure their component matrices.Finally,the resultant cipher image is obtained through the pixel confusion operation,the random cutting operation and filling method,the multi-round row and column diffusion process,respectively.The receiver uses the decryption and IRLS reconstruction algorithm to recover the original image.Several experimental results and security analysis demonstrate that this algorithm has high security and strong robustness.3.A visually secure data hiding in encrypted image algorithm based on compressive sensing and DNA coding is proposed.On the sender side,the embedding process is as follows:Firstly,the original image and two improved chaotic systems are used to generate the chaotic sequences.Secondly,the red,green and blue components are extracted from the secret image,and these components are sparsely represented by KSVD technology.Then,the measurement matrices constructed the chaotic sequences are applied to compress three components for getting the measurements.In what follows,the DNA coding technology and key streams diffusion technology are applied to encrypt the measurements.The final encrypted image is obtained.Finally,the DNA level least significant bit(LSB)dynamic embedding method is used to embed the encrypted image on the carrier image,so as to obtain a visually safe image.Image extraction algorithm is the inverse operation of embedding algorithm.On the receiver side,the embedded image is extracted by the extraction algorithm.Simulation results and performance analysis prove that the proposed method has higher information hiding capacity,higher security and robustness.