Block-Oriented Information Compression

Data base statistics play an important role in conventional information compression. For a large data base, the acquisition of data base statistics becomes a very difficult task. This paper presents a new scheme for information compression that does not use information statistics. Each information block is represented by two sub-blocks called the alphabet and the generator. The alphabet contains the linearly independent elements; the generator is computed through the linear combination of the linearly dependent elements. The total length of these two sub-blocks is generally shorter (never greater) than the original block.