A New Variable-Length Integer Code for Integer Representation and Its Application to Text Compression

Data Compression plays an important role in reducing data storage space in computer memory and in achieving minimum data transmission time in communication networks. There are two types of data compression: lossless and lossy. In lossless data compression, decompression reproduces data that is exactly match the original data and in lossy data compression, the decompression reproduces data which is an approximation of the original data. Variable length integer codes such as Elias Gamma Code, Elias Delta Code, Golomb Code, have been used for data compression (i.e. integer compression, text compression, etc). In this paper, a new variable length integer code is proposed based on radix conversion and it is used with Burrows Wheeler Transform for text data compression. The performance of the proposed code is compared with Elias Gamma Code, Elias Delta Code and Golomb Code. For evaluation, Calgary corpus is used in the experiments, which contains both text file and binary files. Experimental results show that the Fibonacci code gives better compression rate on an average than all other coders and Elias Gamma code gives better compression rate for text files. The other coders perform well for binary files compared to Elias gamma code.