Predykcyjno-substytucyjna metoda bezstratnej kompresji danych

The purpose of this thesis was to define a new method of lossless data compression, combining elements chosen from existing methods of a different kind, and empirical verification whether the solution, based on two methods of which one is fast, and the other effective, will be better than those, when both speed and effectiveness are taken into account.

Existing data compression methods have been examined, looking for ideas which could be included in the new method. The problems of combining two lossless data compression methods have been investigated (switching rules, exclusion rules). By showing the inefficiency of existing methods, the need for a new one has been substantiated. By showing the mutually complementary properties of the predictive and substitutional methods, the choice of the component methods has been substantiated.

The proposed algorithm consists of a mix of existing (range coder, PPMC, though modified according to the author's ideas), and new solutions (fast switching cost evaluation, incremental switching distance encoding, zero-order cache and distance dictionaries). The verification has been conveyed empirically. In its preliminary stage, the algorithm's behavior depending on different values of input parameters has been examined. Also, it has been compared to the predictive component algorithm showing the improvement due to the switching method. Then, the proposed algorithm's results have been compared to the results produced by the best available data compression algorithms, proving the usefulness of the proposed algorithm.