Komprese pomocí algoritmu Context-Mixing

Abstract

Algorithms based on the principle of Context-mixing currently belongs among compressors which achieve the best results. This excellent compression ratio they reach at the expense of his running speed and memory, they need to compress. The principle of these algorithms is the maintenance of several statistical models and appropriate mixing their results so that compression reaching the highest compression ratio. In this diploma thesis we become familiar with the algorithm DiplomaCM which implements appropriate statistical models and exponential context-mixing is particularly noticeable at its higher speeds. The following is the concrete implementation of the algorithm and tests its compression ratio and the compression time compared to existing methods or algorithms. The conclusion also suggest possible improvements and description possibilities of parallel implementation of this algorithm.