Research On Cloud-assisted Watermarking Based On Compressive Sensing

With the rapid development of computer information processing technology and communication technology, large amounts of data will be produced every day. For some resource-limited devices, how to deal with these large amounts of information and provide security protection is a crucial problem. A feasible method is to combine compressive sensing(CS) and Cloud to provide data transmission, storage and management and other functions; and data hiding technology is a great way to provide copyright protection of the data, which realizes the function by hiding some secret information in digital media. In view of the above problems, the main work of this thesis includes the following three aspects:1) The theory of compressive sensing is researched, including sparse signal representation, measurement matrix design and sparse reconstruction algorithms, and some current mainstream reconstruction algorithms are introduced. On the basis of this, the existing cloud-assisted schemes based on compressive sensing have been studied deeply, and the core algorithms have been mastered, which can lay solid foundation for further improvement and design.2) The basic algorithms, classification and main evaluation indexes of data hiding are studied, and the current data hiding schemes based on compressive sensing are researched deeply. Then a double domain watermarking algorithm based on compressive sensing is proposed, in which the data embedding is realized in the compressive sensing domain and the data extracting is realized in the sparse domain. This scheme does not affect the efficiency of transmission and has good robustness. Although it is not completely reversible data hiding algorithm, the accuracy of the data obtained after the watermark extraction can be improved.3) The application background of cloud-assisted is added into this scheme, which is suitable for not only the resource-constrained sampling end, but also the users with limited resources. The algorithm transfers the main computational work to cloud, and provides good security for the data at the same time. Experimental results show that the scheme is suitable for the large-scale data transmission and storage. It can greatly reduce the computation amount of users with cloud-assistance, and can provide good copyright protection of data.