Scientific Data Compression Schemes Based On Preprocessing

In scientific computing fields, with the development of high-performance parallel computer and computing theory, simulations of complex phenomena in full space with high precise and resolution are in sight. At the same time, these applications produce larger volumes of data which often overburden storage and transmission devices. As existing compression algorithms and tools hardly meet the scientific data compression needs, we must design better lossy and lossless compression schemes that typically intend for simulation results.In this dissertation, we propose three new compression methods of scientific data. The first one is a lossless compression method which based on lifted wavelet transformation. The second one is a lossy compression method which based on prediction. The third one is also a lossy compression method which based on multiresolution analysis. Of course, all these methods come out after a learning of entropy and take into account the character of simulate results. The proposed methods provide effective ways for different scientific computing applications. The dissertation is organized with six chapters.Chapter 1 summarized backgrounds and significance of scientific data compression, listed out some main research results together with their character, adaptive area and shortages.Chapter 2 gave a presentation of basic theory related to data compression, listed out several encode algorithms, surveyed some problems and approaches that scientific data compression must face.Chapter 3 introduced the theory of lifted wavelet transformation, proposed a...