QLFC - a compression algorithm using the Burrows-Wheeler transform

Summary form only given. In this paper, we propose a novel approach for the second step of the Burrows-Wheeler compression algorithm, based on the idea that the probabilities of events are not continuous valued, but are rather quantized with respect to a specific class of base functions. The first pass of encoding transforms the input sequence x into sequence x˜. The second pass models and codes x˜using entropy coding. The entropy decoding, modeling, and context updating for decoding x˜ are the same as the ones used for encoding. We have proved that the quantized local frequency transform is optimal in the case of binary and ternary alphabet memoryless sources, showing that x and x˜ have the same entropy; for larger alphabets, we verified this by simulation.