A multi-block data compression method based on arithmetic codingfor a library database

摘要

Statistical analysis of the distributional and correlationalproperties of the characters in a large library data file with formattedrecords leads to the employment of a number of cumulative probabilitydistributions to encode characters in various fields of each record. Theperformance of the proposed technique is compared with those of two-passand adaptive character-based arithmetic coding methods. Experimentalresults show that the proposed technique yields a 13.6 to 13.9% increasein compression for a file in the Sydney Micro Library System