Image Compression-encryption Hybrid Algorithms Based On Key-controlled Measurement Matrix In Compressive Sensing

In recent years, compressive sensing has been attracting many attentions as anew signal sampling theory. And many image encryption algorithms based oncompressive sensing have been proposed. The principle of compressive sensing,random pixel exchange, Arnold transform are introduced in the dissertation. And fourimage compression-encryption algorithms based on compressive sensing areproposed by combining the random pixel exchange, Arnold transform or indexscrambling. The main results are as follows:An image compression-encryption hybrid algorithm based on key-controlledmeasurement matrix in compressive sensing and random pixel exchange is proposed.The original image is divided into four blocks to encrypt and compress. Then thepixels of the adjacent encrypted and compressed blocks are exchanged by the randompixel exchange algorithm with random matrices. The measurement matrices incompressive sensing are constructed by utilizing the circulant matrices andcontrolling the original row vectors of the circulant matrices with Logistic map. Andby binding the random matrices with the measurement matrices, the security ofsystem is enhanced. The security of the proposed algorithm is verified by theexperimental results.A new image compression-encryption algorithm based on compressive sensingis proposed, which is practical and can accomplish image encryption and compressionsimultaneously. The partial Hadamard matrix is adopted as measurement matrix,which is controlled by chaos map. The measurement is scrambled by indexscrambling. The proposed algorithm is practical, sensitive to the keys and can resistvarious attacks and reduce the burden of transferring keys compared with the schemesusing the whole measurement matrix as key. Simulation results verify the validity andreliability of the proposed algorithm.With the orthogonality of Hadamard matrix, an image compression-encryptionalgorithm based on2-D compressive sensing is proposed, which can solve theproblem that treating the measurement as the output cannot resist the chosen-plaintext attack. The measurements are performed in two orthogonal directions and themeasurement matrices are constructed as partial Hadamard matrices, which arecontrolled by chaos map. The proposed algorithm is sensitive to the keys and canresist various attacks. Simulation results verify the validity and reliability of theproposed algorithm.A new color image encryption algorithm combining compressive sensing withArnold transform is proposed, which can encrypt the color image into a gray image.Since the dimensional reduction and random projection of compressive sensing, it isutilized to encrypt and compress the three color components of color imagesimultaneously. The dimensions of three encrypted and compressed color componentsare smaller than the original image, thus they can be grouped into a gray image, thenthe gray image is scrambled by Arnold transform to enhance the security. Theproposed algorithm can also be applied in multiple-image encryption. The simulationshows the validity and the reliability of the proposed algorithm.